a Sey *! aoe : ale Ve, ! v6G OLREX eer. SMITHSONIAN INSTITUTION UNITED STATES NATIONAL MUSEUM Bulletin 61] VARIATIONS AND GENETIC RELATIONSHIPS OF THE GARTER-SNAKES | BY ALEXANDER G. RUTHVEN Curator of the University Museum, University of Michigan, Ann Arbor ae TR ONORs ITV YNGTONS WASHINGTON GOVERNMENT PRINTING OFFICE 1908 ADVERTISEMENT. The scientific publications of the National Museum consist of two series—the Bulletin and the Proceedings. The Bulletin, publication of which was begun in 1875, is a series of more or less extensive works intended to illustrate the collections of the United States National Museum and, with the exception noted below, is issued separately. These bulletins are monographic in scope and are devoted principally to the discussion of large zoological and botanical groups, faunas and floras, bibliographies of eminent natural- ists, reports of expeditions, etc. They are usually of octavo size, although a quarto form, known as the Special Bulletin, has been adopted in a few instances in which a larger page was deemed indis- pensable. This work forms No. 61 of the Bulletin series. Since 1902 the volumes of the series known as “Contributions from the National Herbarium,” and containing papers relating to the botan- ical collections of the Museum, have been published as bulletins. The Proceedings, the first volume of which was issued in 1878, are intended as a medium of publication of brief original papers based on the collections of the National Museum, and setting forth newly acquired facts in biology, anthropology, and geology derived there- from, or containing descriptions of new forms and revisions of limited groups. A volume is issued annually, or oftener, for distribution to libraries and scientific establishments, and in view of the impor- tance of the more prompt dissemination of new facts a limited edition of each paper is printed in pamphlet form in advance. RicHArD RATHBUN, Assistant Secretary, Smithsonian Institution, in charge of the United States National Museum. Wasuineton. U.S. A., June 15, 1908. III TABLE OF CONTENTS. Page. “TET ie CO COTS STS ee AN Vv Sai ei Umnim PPA TONS Gtr ee Soak Ne exe oda x -in\cre.n ocho ree Shee ade s IX PRL ON CLEARED Pitot es Sees eee SE Sk es ee We aE 1 iilie taxonomiy.ol te Partersnakes..2 2. 2...2.0.2 2 -....2eccsim shee cee eee Sa 3 Pleremt Pawe OMi Mer eOnta it. Soe ek ee tee oe ade ole 3 Methods employed...............-.- Sete pete ahs aaa AS AG ae Mr: ee 4 NLS AET O10 oy Nia aTi a2 6 | Mi Seaplane et AS RE RIN eV Sl, SN 7 PACINO LGM OMEN Ia seco tee cet m2 Men Bala Oe eM AY RI Ste Jeske es 7 SEDO TSO) 0 TSIEN, thet ene Oo a 8 BEG ENDO Ml octane at see tn re AP nh A ESS ee No ae ws ue oe tee 8 Srennbe cinta Wena ey pene erate AE NEN ok ue SN ee 8 ARN ee eee oe Rye eT Sa Ee aa ate eas sete oe Ne Aa veine Je ae 10 Ee eae Pee eae, stony ha St es RSS De ee oe Eh ace 12 Ea bitaandanalitamnelationsa - 5-265 .cc.262.c eel 36.02 tla 12 WTEC) fe SE SOLS ROE es ep re RE me ee ne ae ene 15 SENATE LON A TTT STU G21 ES (0) 0 ea en OD mC Ik 16 Variation in muimber of dorsal scale rows...-.../...-.----+--+-22 22-2 16 Vanation in number: ot talbial plates. ..2.--4:- 22.42 J2.2-.- 222. See 22 Variation in number of ventral and subcaudal plates, and in propor- mona tcatail-Lenpinietiecstte ss. aU eee. oN owe eect bee ee 31 Variation in number of preoculars, and in arrangement of lateral spots. 32 Verio ptomneneripeser no tort ee MSs tin oe coe sk SL Gere Se oe 33 Wemriniinl ann GOlLar yee pee 2 te 2s Pg es Pea IS tae 37 itretingn eToupS Gh Parter-snakes ors. yl eos l so. ge ee ee A ech 39 The Radix group (megalops, marcianus, radix, and butleri) .................- 44 eS Ornette See ey ees dene ee ake 44 HUES GUT E DONS iso ho epee | ee a a eee ee ee ae A a 44 Papiieranadviabitat Telations 222s 6.20220. ..2 SERIE fetes Be Re ee Sg a . Thamnophis butleri, showing the last stage in the loss of the sixth supra- labial—the fusion of the reduced scute with the fifth.............. . Distribution of Thamnophis megalops, as indicated by the locahiig TRE COIAO [Stan ees eRe cat Need Ar OR dons ph a Dos Sat ae eee ee . Diagram showing the variation in the dorsal scale formula in Tham- TIGh PLES STE OO PSE Sein pects aoe Ls A ee ee Caer. ee eS Sef . Diagram showing the variation in the number of supralabials in Tham- peued eS A REE ALONG! tees ee te eee. MMe eee Oe eT Oya ba zoo. . Diagram showing the variation in the number of infralabials in Tham- MIS COAL O Pea be sae oc oS ee ee es Seto doe ea -. - - . Diagram showing the variation in the number of ventral scutes in ; ees shomaee ee variation in the number of subcaudal scutes in Mingmmeghis merdlOps-e ees eaes odes See Se See fi oe wre . Diagram showing the variation in the proportionate tail-length in (Mirae op Mins AODS tart cook aie cei tee hee So. 6 ce ace eee . Santa Cruz River at Tucson, Arizona. Thamnophis marcianus has been found here. (From Ruthven’s Reptiles and Amphibians from ING walle COLA Gwe ATIZONA) me = somes tise ee cteccs a5 Ses eee eee wt = Page. 27 Fic. 2 ioe) oo 42. 43. 44, 45. 48. LIST OF ILLUSTRATIONS. The natural divisions (environmental complexes) of the Greater Texas Region. (From Hill, Physical Geography of the Texas Region)... . Distribution of Thamnophis marcianus, as indicated by the locality TOCOLGS. «a= b-orele acc & o/c)5 ite Sale SSF ere ee rn Habitat of Thamnophis radix and T. sirtalis parietalis. Slough (outlet to. Elbow Lake), Clay Coumty,clowass. 22. See ee ee Habitat of Thamnophis radix and T. sirtalis parietalis. Morainic hills in Clay County, lowa 222228 ee as ee oe oe ee ee Distribution of Thamnophis radix, as indicated by the locality records. Diagram showing the variation in the dorsal scale formula in Tham- nophis radix. . -.. SNe eee ee tes ee aie eee eee aes Diagram showing the variation in the number of supralabials in Tham- nophis radix.......d2.222a, Co ee es epee eye a Diagram showing the variation in the number of infralabials in Tham- nophis radi€:+is2.2 $28te eas Soe Ee eee eee ee Habitat of Thamnophis butleri. Creek at Lima Center, Washtenaw County, Michigam 22 ecautie nn nee tone neo = ce re ene ee ; Distribution of Thamnophis butleri, as indicated by the locality records. Diagram showing the variation in the dorsal scale formula in Tham- nophis‘butlery. 52. 4 Dai oles ie Yt Pe ee eee Diagram showing the variation in the number of supralabials in Tham- nophis: brblene £2 5: oh .98 eso 4 SSL Th A a ky Diagram showing the variation in the number of infralabials in Tham- Distribution of Thamnophis sauritus proximus, as indicated by the localityeiecords:ws 2k. r oe ae noe ht RE ere ey Ae Rene Diagram showing the variation in the number of supralabials in Tham- nophissagriias proxies, 92. eB Pale ee ae Diagram showing the variation in the number of ventral scutes in Thamnophis sauritus proximiugss.*-62425 52 See tee eee Diagram showing the variation in the proportionate tail length in Thamno pis sanritus proximate. 2.52: Sea Fee es ee ee ee Diagram showing the variation in the number of subcaudal scutes in Thamnophis: sauritus}prommiis. <3. S-seaes) tse eee eee Distribution of Thamnophis sauritus sackeni, as indicated by the locality records. (7 Ree ys Ee: oe a eee ee Diagram showing the variation in the number of supralabials in Tham- nophissauxcibusisackent ee. oo... Shack Shee eee nee Diagram showing the variation in the number of ventral scutes in Thamnophis:saurittisssackemt..- 2.0 224-5 see ee eee Diagram showing the variation in the proportionate tail length in ‘Thamnophis sauritus sackemi:?.. 3-252 a4e eee ee ee Diagram showing the variation in the number of subcaudal scutes in Thamnophis sauritus'sackeni..<.¢ se S-. -=.) 222d See ee Distribution of Thamnophis sauritus, as indicated by the locality FOCOEdSS2.25 - wo eked Bese oe he ee Diagram showing the variation in the number of supralabials in Tham- NOPHIS salirituss 2. - gies. Mee ahs SS ee er ree ee ees Diagram showing the variation in the number of infralabials in Tham- nophis SaiEWMss 2 2.275 Met ee woe oten | Ree ee eo ee a ie ee Diagram showing the variation in the number of ventral scutes in Thamnophissaumibiga eee SeUe See SU Soe ee ee ee eee iat Page. Fic. 49. 62. LIST OF ILLUSTRATIONS. Diagram showing the variation in the proportionate tail length in SP omat io late SAMIPIOVIS=: <2 pee eee eee ee a. ee oe Diagram showing the variation in the number of subcaudal scutes in ise UIE cAUInIiN 55:12 82 seo eee ks t a hcbreh..... Phylogenetic development of the Sauritus group..............-..--. Distribution of Thamnophis angustirostris, as indicated by the locality RE DLC ES A okey SA: aL ee Re as Sy oa) 5G ee a Distribution of Thamnophis angustirostris melanogaster, as indicated ba atiesloea hii, NeCOnds. 4: -2hel S24. x \s soot ee ee PE) Distribution of Thamnophisscalaris,as indicated by the locality records. Habitat of Thamnophis hammondi. Stream at Baldwin’s ranch, about 20 miles east of Los Angeles, California. (Photograph by \SUTHINE eahe) breg 0 i Coa) Renee ete ge” eee TN a Distribution of Thamnophis hammondi, as indicated by the locality ig 200 ule See oP een, Miah or eae a Rik Se Pal a ge eee MRE iol POLAR oe Diagram showing the variation in the number of preoculars in Tham- SO TLNUS PAA GES eh TSN pac lo ar soe ae eee A ID ee he ute Habitat of Thamnophis ordinoides elegans. Klamath marshes, south- ern Oregon. (Photograph by Finley and Bohlman)............... Distribution of Thamnophis ordinoides elegans, as indicated by the scriya record aes casae te tet ae lee Cie SE ao ee Diagram showing the variation in the dorsal scale formula in Tham- Up MAS ONGUN Gites Cleoamn es. 2 os. 56 eee. Abele a os lk Diagram showing the variation in the number of preoculars in Tham- TopAnis OndinOrderie eGANsys 654. sfa. chelsea ss ee ee le Distribution of Thamnophis ordinoides, as indicated by the locality BS COL SNe rcicst Heme ran oe Aue Re ke ee eS ee SINS Diagram showing the variation in the dorsal scale formula in Tham- TAOS OLMIMIOLAeS weet te Saar eo ee ee et he ee ee Diagram showing the variation in the number of supralabials in Mhammophis'ordimondess:c.).: svi. 22.01 258205. ke eee Diagram showing the variation in the number of infralabials in Tham- OVO) AUAShCOr Gl TOTO KE (=i) 5, eee Ren Sq eae ey Set a WC ero Ver Diagram showing the variation in the number of ventral scutes in Ahi asm phd ordi Oudes: 2s 2= ste. Sere on ete Diagram showing the variation in the number of subcaudal scutes in prmneuitis mcdingides \w.cls2. ek ee LR ee Diagram showing the variation in the number of preoculars in Tham- NG PHISOrAMRO eR: Bee Saree et ee ey Le le ob ee Phylogenetic development of the Elegans TOUS sects ttn see ears ee ee Distribution of Thamnophis eques, as indicated by the locality records. Diagram showing the variation in the dorsal scale formula in Tham- DRG ee GMOS eee ensue ee eR OO uy Pee Ny | Diagram showing the variation in the number of supralabials in Tham- [COT OIOIS) Sit UES oh sea Ne ae a ae Distribution of Thamnophis sirtalis parietalis, as indicated by the ue CallU ACEO TORE: noe See a oe A Variation in the number of supralabials in 20 specimens of Tham- nophis sirtalis parietalis from Montana....................-.-..--.- Variation in the number of infralabials in 20 specimens of Thamnophis sirt talis Teese from Mauteng 174 LIST OF ILLUSTRATIONS. nophis sirtalis from southeastern Michigan....................._... 182 Variation in the number of infralabials in 113 specimens of Thamnophis sirtalis from southeastern Michigan............................... 183 Diagram showing the variation in the number of ventral scutes in Thammophis sirtalis.._-pessess ee eee ee a ee 183 2. Phylogenetic development of the Sirtalis Prolipers: tre Seay eee 186 {ha we 1 oe = PLE. BULLETIN NO. 61 U. S. NATIONAL MUSEUM ‘(NOD3YO ‘3XV7 31NL) SNVD3713 SSCIONIGHO SIHDONWVH VARIATIONS AND GENETIC RELATIONSHIPS OF THE GARTER-SNAKES. By ALEXANDER G. RUTHVEN, Curator of the University Museum, University of Michigan, Ann Arbor. INTRODUCTION. In these times, when there is so great activity in evolutionary research along all lines of biological work, it is important that sys- tematic studies be directed so as to throw light on the problems involved in the origin of species. An examination of the actual sytematic work that has been done, however, can not fail to reveal the fact that only a comparatively small amount of it has been done with this end in view. ‘This is especially true in the case of reptiles, and more particularly in the order Serpentes, where but little work has been contributed to this phase of the subject. Without doubt one reason for this is to be found in the fact that the snakes are a most perplexing group systematically, owing to the difficulty in finding characters that are capable of defining natural groups. For this reason but. a comparatively small amount of work has been done on the group. As Baird said in 1853 (1853, p. v.), ‘‘Systematic workers all carefully avoid the subject of Ophidians, each waiting - for the others to make the first step;’’ and although in the fifty years that have elapsed since this statement was made considerable work has been done upon the order, and a multitude of forms have been described as the result of the careful studies of Baird, Stejneger, Cope, Garman, Brown, Van Denburgh, and others, only a very few con- clusions have been reached as to the origin and descent of the forms. This barrenness of general results can not be entirely attributed to a lack of facts, but must be inherent, to some degree at least, in the methods employed. An examination of the systematic work that has been done upon the snakes shows, as might be expected, that it is largely analytical in its nature, being for the most part descriptive of the existing diversities. That this kind of work is of value can not be questioned, for, as has been well said by Tutt (1896, p. 6): ‘‘The species describer, if he does his work intelligently and carefully, is giving the evolutionist the exact material on which alone il 2 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. any stable conclusions can be drawn with precision, and, if his work~ is thus valuable, we must still recognize, nay, welcome, those who give their best powers to the unraveling of the species in their multi- tudinous forms, since these form the basis of all advanced evolution- ary study.” While the value of the analytical method must, there- fore, be admitted, its limitations should be clearly seen. Needless to say it is not a knowledge of the present conditions alone that is necessary, Which is all that the application of this method can yield, but a knowledge of the processes that have brought them about, for systematic work can only become a true science when it seeks to formulate the laws involved in the history of the present forms. The history of a particular form can not be worked out by determin- ing its characters alone, but its affinities must be sought for, by determining its similarities with other forms, and the factors which influence it, before its derivation can be discovered. After analysis, therefore, as has been said, comes the need of a larger synthesis. This argument is often verbally granted by systematists, who, how- ever, still adhere to the analytical method with the plea that a suf- ficient body of facts has not as yet been accumulated upon which to base general conclusions. This may probably be said of snakes with more truth than for many other groups, but, when it is noted that the work that has been done offers very little material upon which to base such general conclusions, it would seem that a few generaliza- tions based upon such limited material as is now available might in many cases go far toward directing the course of future analytical work toward better results. In the words of Meldola (1896, p. 7), ‘However large the number of facts, and however cautious or con- servative the worker may be, it is an established doctrine, taught by the whole history of science, that real progress begins when we go to seek for facts armed with at least the suggestion of a principle if not with a complete theory based on facts already accumulated by observation or experiment.” Furthermore, it seems to be evident that any comprehensive attempt to do synthetical systematic work must be made upon a geographic basis, for the interrelations between the organisms and their environments are such that the history of the present forms is involved with the history of the conditions with which they are associ- ated, and affinities can only be sought in the light of geographic probability. As Adams (1902a, 356) puts it, ‘‘It should be apparent that in the past history of a region the conditions and highways must be taken into consideration if we wish to understand the origin and migrations of the diverse elements which enter into the fauna and flora of any given region.”’ As a study of the interrelations between organisms and the natural environment, geographic distribution VARIATIONS OF GARTER-SNAKES. 3 attains to a wider significance than it has previously enjoyed. Hith- erto, as the result of the application of analytical methods to tax- onomy, it has been largely static in its nature, as it concerned itself principally with the determination of the present distribution of the forms as if this arrangement were fixed, thus affording only a descrip- tion of the present conditions. In connection with the synthetic method, however, it is more comprehensive and essentially dynamic in its nature, for it not only takes note of the present distribution of the forms but at the same time recognizes the close relations that exist between this arrangement and the environment, and its conse- quent unstability,and by the determination of the laws of change furnishes a powerful aid to the interpretation of the past history of the present forms, without a knowledge of which it is impossible to explain the present conditions. THE TAXONOMY OF THE GARTER-SNAKES. PRESENT STATUS OF THE GENUS. As an example of the confusion into which a group may be thrown by systematists who see in every association of traits a species, prob- ably no better could be found than that of the common garter-snakes of North and Middle America, composing the genus Thamnophis Fitzinger. This genus has long stood in the minds of herpetologists as a synonym for chaos, and has been a source of trouble to those who attempted to compile local lists, for specimens may be found in almost any locality which can not be referred exactly to any described form, or doubtfully to two or three. The result has been that in the attempt to express these differences the method of analysis has been carried to the extreme, and something like sixty-five forms have been described. This has not solved the problem, however, for, owing to the extent of variation and the fact that descriptions have been based on differ- ent combinations of traits without an adequate knowledge of the variations, it has been practically impossible to define accurately any of these forms. In nearly every species the characters often seemingly erade off in different directions independently of one another, and it has happened more than once that a diagnosis has been subsequently shifted from the form represented by the type-specimen to an entirely different one, by the author of the species himself. Anything approach- ing a natural key was not to be thought of, and a worker was fortunate if he could devise an artificial one that would sufficiently define the forms in particular regions so that they could be recognized. It is an instance where the insufficiencies of the methods applied have been thrown prominently into the foreground by the nature of the material, 4 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. for a widely ranging and variable group with all the diversities brought about by environmental and racial influences doesnot furnish the most favorable material for the application of purely analytical methods. An examination of the most comprehensive work that has been done on the garter-snakes will show this. In his ‘‘Crocodilians, Lizards, and Snakes of North America” (1900), Cope has described the diversities in an exhaustive way, recognizing in all forty-six forms. Diversity is all that this work reveals, however, and a more minute splitting of the forms would only increase the complexity. Here one is shown the actual living varieties as they exist in North America to-day; the present conditions with all of their anomalies and apparent contradictions, with no key to possible relationships.¢ What is needed, it seems to me, is not more analysis but a greater knowledge of the affinities of the forms, so that the continuity of genetic relationships underlying the present diversities can be grasped. METHODS EMPLOYED. Three steps are necessary to determine the genetic relationships and simplify Cope’s elaborate arrangement of the group: (1) The value of the characters must be determined; (2) the geographic prob- abilities must be utilized; (8) similarities and intergradations must be sought. It may seem superfluous to insist upon the determination of the value of the characters used in systematic work on this group, for ‘“‘good”’ characters, 1. e., those which are constant within particular groups, are sought after by most systematists in forming their taxo- nomic groups, but in the sense in which it is here used this rule means more than this, in that it also requires the discovery of the significance of the variations. In the work that has been done, as has been said previously, when- ever a more or less stable combination of traits has been observed it has been described as a distinct form, and this with utter disregard as to how the distinctive characters may have been derived. For instance, elegans is separated from radix by the position of the lateral stripe, from parietalis by the smaller number of dorsal scale rows; sirtalis is separated from radix by the position of the lateral stripe and by a difference of two rews of dorsal scales, from butleri apparently only by the width of the lateral stripe. In these cases it has been sufficient to ascertain that these characters are little variable within the form, but what relations can be worked out from such facts until «Tt is true that twenty-seven of the described forms are considered subspecies, and a diagram of affinities has been worked out, but the latter is mostly hypothetical, while most of the subspecies are either based merely upon individual variation or are erroneously arranged. VARIATIONS OF GARTER-SNAKES. 5 it is ascertained to what extent these characters are subject to modi- fications? Only when this is known can we hope to accomplish much in our attempt to discover the origin of the present forms. This is very well illustrated in the synthetic taxonomic work that has been attempted. Several efforts have been made to combine the described forms into natural groups, and Stejneger, Brown, and Van Denburgh * have, in various papers, endeavored to establish the proper status of the different forms. The most comprehensive essay is that of Brown (1901) in his ‘‘Review of the Genera and Species of American Snakes North of Mexico,” which appeared in 1901, and which may be considered as a reaction from the extreme position taken by Cope (1892) in his paper ‘‘A Critical Review of the Characters and Variations of the Snakes of North America.’’ In this work Brown has attempted to unite the different forms into related groups, and, as will be shown later, has succeeded very well in some instances. But commendable as are his efforts, the methods employed, it seems to me, are unfortunate. For example, sirtalis, leptocephalus (ordi- noides), butleri, parietalis, and pickeringi (concinnus) are united in one group on the basis of similarity in the scutellation without regard as to whether or not the characters included under this head are influ- enced by definite variation, and until this is determined there may be another explanation for similarity in scutellation than community of origin, namely, parallel development. That there is convergence among. the garter-snakes is illustrated by the fact that so high an authority as Boulenger (1893, 418) gave it as his opinion that the type of brachystoma (butleri) was a specimen of leptocephalus (ordinoides), incorrectly labeled as to locality. As Cope subsequently showed, there is no doubt as to the correctness of the locality, so that it is highly improbable that the two forms are directly related, and yet the parallelism is so close that if unlabeled as to locality it is often very difficult to distinguish between speci- mens of the two forms. (See also p. 188.) It is essential, therefore, to determine first of all along what lines the present combinations of characters may have developed, in order that there may be no confusion by parallelism. Another prime objection to the synthetic work that has been attempted is that the geographic probabilities have been so largely dis- regarded. A knowledge of the extent of the variation in any form affords little evidence of affinities if the geographic trends are ignored @ Garman’s arrangement (1883) is in many respects the most comprehensive, as he considers the group in detail. There is, however, so little basis given for the grouping which he has adopted, and so many manifest errors, that his synopsis must be consid- ered rather as an effort to reduce the number of species than as a rational attempt to establish natural affinities. 33553—Bull. 61—08——2 6 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. as they always have been in this genus. And again while the law that the direct relative of any form will generally be found in a neighboring environment (see p. 192) may not be universal among the vertebrates the evidence at the present time shows that it is valid in a great number of cases, so that one must be very cautious of this poimt in grouping the forms of any group. Cope in his general work recognized five varieties of sirtalis east of the Mississippi River, and yet the range of these forms could not be defined relative to each other, as they were reported from various parts of the same region and in association with numerous other varieties of the same stock, in utter defiance of the above law, but as most of these forms were based on shght variations and have been mostly dropped, they do not merit detailed discussion. The affinities that have been claimed for recognizedly distinct forms, however, is open to the same criticism. For example, if ordinoides and concinnus are both derived from parietalis, one must invoke the aid of some form of physiological isolation to account for their present distinctness, for they occupy the same region and environment, and concinnus is not considered far enough separated from the parent stock to be classed as a distinct species. Similarly also the derivation of radix from sirtalis is different from what one might expect from their geographic location, for radiz lies entirely within the range of sirtalis and parietalis. Thus geographic probability must also be observed in tracing genetic lines if we would avoid the grouping of forms of diverse origin, which by their similarity might reasonably be adjudged to be related. Having determined the significance of the variations, and by observing the geographic probabilities, considerable advance may be made toward determining genetic relationships, and it only remains to look for similarities and intergradations. For example, we may thus have determined that the smaller scutellation of ordinoides may have been derived by dwarfing from some other form with a larger number of scales, and that it is possible, geographically, that it may have been derived from the snakes of southern California or from those in the Great Basin. It is not enough to say that the present combination of traits which we call ordinoides is constant, but we must cast about for similarities with neighboring forms in order to determine its origin if we are to avoid the multitudinous varieties of systematists whose vision is blurred by the present divergences. The species-describer searches for pure types which are found only away from the geo- graphical meeting places, but if we wish to discover relationships we must not avoid the boundary lines, but seek them. If we search for similarities in the light of the influence of modifying factors on the characters, and with an eye to geographic probabilities, we should be able to see back of the present diversities, grasp the natural affinities, and come in touch with the questions of origin and descent. It was VARIATIONS OF GARTER-SNAKES. cf in the belief that the application of these methods to the group of garter-snakes would lead to such results, and thus greatly simplify our present knowledge, that this work was undertaken. MATERIAL EXAMINED. The conclusions reached in this paper are based upon a detailed examination of about three thousand specimens gathered from nearly every part of the known range of the genus,” so that I believe my knowledge of the variations is well founded so far as it goes. It must be acknowledged, however, that much remains to be done, for, owing to the fact that it has been found impossible to obtain extensive series from many localities, that entire forms are at present known from but few specimens, while in the case of some well-known forms specimens are unavailable from critical localities, it is impossible, in some instances, to do more at present than point out the probable relations, leaving them to be substantiated by detailed work when specimens shall be available. One objection to the material that has been used is that it has been largely alcoholic, and it must be admitted that, as some of it has been in preservation for fifty years, it has not always been all that could be desired. This has been more of a limitation to the work, however, than a source of error, for the principal characters employed in the systematic work (scutellation, position of stripes, and tail length) can not become greatly modified in this way, and, although the color may become very much faded, the color pattern may in most cases still be made out. A far more important objection to the material is that many of the specimens have but general locality labels, and throughout the pres- ent work it has been continually impressed upon the writer that too much attention can not be paid by collectors to the securing of detailed locality and habitat data. Specimens that are not labeled accurately, at least as to locality, are almost valueless in this work, while if habitat data were only available in each case much light would be thrown upon the explanation of the present distribution. ACKNOWLEDGMENTS. ~ In the prosecution of the research upon which this paper is based I have become indebted to a number of persons without whose aid little could have been accomplished. | I wish first of all to acknowledge my indebtedness to Mr. Charles C. Adams, under whose direction the work has been carried on, for his @The specimens examined represent every described form, with the exception of the Hutaenia praeocularis Bocourt (Le Natur, 1892,-p. 278), from Belize, British Honduras. I am not certain of the identity of this form. 8 ‘BULLETIN 61, UNITED STATES NATIONAL MUSEUM. continued interest in its progress, and for many helpful suggestions and criticisms, particularly upon the general method of approach. T am also areatly indebted to Dr. Leonhard Stejneger for the privi- lege of studying the material in the United States National Museum; to Mr. A. E. Brown for the opportunity of examining the specimens in the Academy of Natural Sciences of Philadelphia, and to Messrs. Whitmer Stone and James A. G. Rehn for assistance in doing so. I am particularly indebted to the Field Museum of Natural His- tory in Chicago, through Dr. 8. E. Meek, for the loan of their excel- lent series of specimens, which has furnished the solution for several knotty problems that otherwise must have remained obscure. For the loan of the material in the American Museum of Natural History I am indebted to the Director, Dr. H. C. Bumpus. My thanks are also due to Dr. Raymond Pearl, Agricultural Experi- ment Station, Orono, Maine, who suggested the method employed in plotting the variations in scutellation in the different forms. Most students of systematic zoology avoid statistical methods, which is unfortunate, as a graphic description is always much clearer than a written one. Simple diagrams, similar to the ones used in this paper, are easy to prepare, and summarize in the clearest and most concise manner the geographic variations in a form, and should be em- ployed whenever possible. It may be well to say here that in these diagrams < denotes the arithmetical mean, the heavy lines the range of variations, O the mid-point of the range (the average of the two extreme indie and the numerals in the Teena the number of specimens; also that the stars on the maps do not represent the exact locality from which specimens have been examined. IT am also under obligations to many persons for aid in securing living specimens from localities from which records were desirable. GENUS THAMNOPHIS. DESCRIPTION. Scutellation.—The genus Thamnophis of Fitzinger (1843, 26), established upon the Coluber saurita of Linneus (1766, 385), and later defined as Lutaenia by Baird and Girard (1853, 24), belongs to the family Natricidae, and differs from its nearest American relative, Natriz, by the absence of scale pits and the presence of an undivided anal plate. The cephalic plates are normal, consisting of paired internasals, prefrontals, occipitals, and parietals, and a single frontal. A loreal plate is present; the rostral normal in form and the nasals divided, the nostril being between them. There is usually a single preocular on either side, but in three forms (elegans, ordinoides, and hammond) VARIATIONS OF GARTER-SNAKES. ) there are frequently two, in two others (angustirostris and melanogas- ter) mostly two, and in one (angustirostris) frequently three. The postoculars are most often three in number, occasionally varying individually to two or four, in one form (melanogaster) being very frequently two. The temporal plates are ordinarily one fol- lowed by two or three, but in one form (butleri) there is frequently but a single plate in the second row; occasionally the first temporal is divided, and frequently there are four scales in the second row, but these are individual variations. The supralabials are 6, 7, 8, or 9 on either side, the eye resting upon the third and fourth, or fourth and fifth, except in angustirostris in which the lower postocular is pro- longed forward beneath the eye to separate the orbit from all but one supralabial. There is a triangular mental plate at the symphysis of the, lower jaw, and posteriorly from this the infralabial plates, 8, 9, 10, or 11 (very rarely 12) in number, extend to the posterior angle of the mouth, being separated, except the first pair, which are prolonged inwardly to meet on a median line, by two pair of chin shields. The anterior pair of chin shields is usually the shorter, but in two forms (ordinovdes and elegans) they are quite constantly equal or longer than the posterior pair. The maxillary teeth are apparently as in the other forms in the family, 1. e., rather abruptly longer near the posterior end of the maxillary bone than anteriorly. The genus Atomarchus Cope (Stypo- cemus, Chilopoma) was based on forms in which the maxillary teeth were equal throughout (melanogaster and angustirostris), but, as later stated by Cope, the excess in the length of the posterior teeth is so small in many specimens of other species that it is impossible to distinguish them from apparently isodont specimens of these forms, while we have, moreover, seen specimens of melanogaster in which the posterior teeth were nearly, if not quite, as elongated as in any form in the genus. On the body there are 17 to 23 longitudinal rows of dorsal scales; these being normally arranged in an even number of rows on either side of a median dorsal series. The scales are all keeled with the exception of those of the first row, which are usually smooth or but weakly keeled; in one instance (melanogaster) those of the second row also tend to become smooth, the keels being generally weaker than those of the rows above. The abdominal plates consist of a single series of large transverse scutes that vary in number both individually and racially, the ex- tremes for the genus being about 132 to 180 from chin to anus. The anal plate is almost invariably entire, while the subcaudal plates, which like the abdominal plates are also variable in num- 10 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. ber, consist of a double series of scutes about 49 to 134 in number from the vent to the tip of the tail. Color.—The color pattern may be described in general as three light stripes upon a darker ground, and, although there is consider- able detail and variation in color, this description will be found to hold in a general way for the genus. The ground color on the scales above the lateral stripe varies, as a rule, from dark yellow at one extreme through the olives to black at the other, and usually, when light enough to show it, exhibits longitudinal rows of black spots. Where best developed (radix) these spots are arranged in three longitudinal series on either side, one below the lateral stripe and two above, the individual spots of each series alternating with those of the adjacent rows. The arrangement is not always so regular, however, for from this ideal condition, so to speak, variations occur in various directions. In the greater number of instances the spots of the first row are broken up and do not form a definite series, while in many cases also the spots of the second and third rows are interrupted by the ground color along the keels of the scales. Frequently the spots of all three rows tend to fuse with those of adjacent rows and form transverse bars or blotches, and in one form (scalaris) this is the pattern throughout the length of the body. Occasionally the spots retreat from the scales, becoming reduced to narrow broken lines along the stripes, and in one group (Sauritus) this is the characteristic pattern. In other forms the scales between the spots are occasionally red. In all of the forms exhibiting one of these styles of coloration the spots may be entirely obscured by the darkening of the ground color, which in some forms seems to be the typical condition. . When the skin is stretched the scales are drawn apart, and it is seen that the lateral spots on the scales represent corresponding series on the skin. Those beneath the first row are difficult to ob- serve, as the skin is here apparently less distensible; they are evi- dently, however, like their representatives on the first row of scales in that they are more subject to irregularities, and are seldom to be made out as a definite series. Above the lateral stripe the second and third rows are usually distinct on the skin, whether or not the spots on the scales are broken up or obscured by the ground color. They also show a tendency to fuse, however, and this fusion may either take place by the general fusion of the spots, leaving only scattered light patches to indicate the interspaces; by transverse fusion into bars; by the fusion of the upper row, which may involve also part or all of the second row, the interspaces being limited to the vicinity of the lateral stripe; or more rarely the two series fuse as two longitudinal bands. VARIATIONS OF GARTER-SNAKES. 14 When the spots on the skin are generally fused, those on the scales are either wanting or represented in narrow broken bands along the stripes, as mentioned above. When, however, the latter show a tendency to form cross-bars, the spots on the skin exhibit the same arrangement. Practically always, however, the first spot of the two rows above the lateral stripe fuses with its neighbor to form a large nuchal blotch, which may or may not be distinct on the scales. These blotches may cross the lateral stripe and fuse with the cor- responding spot on the first row of scales, and in one section of the genus (the Sirtalis group) are frequently followed by one to five similar blotches which are likewise formed by the fusion of the spots of the second ‘and third rows. The color of the skin in the interspaces is usually white or very light bluish or greenish, but it may also be yellowish or red, and the color of the interspaces usually encroaches to a varying degree upon the involved scales. The stripes are typically three in number, consisting of a dorsal which involves the median and usually a varying amount of the adja- cent rows, and a lateral on either side situated on the second and third, third, third and fourth, or second, third, and fourth rows from the belly. The color of these stripes is usually brighter than the ground color, being white, various shades of blue, green, or yellow, or occa- sionally red. The laterals, while frequently narrow, usually involve two rows of scales. In one form (butleri) they cover part of three rows, while in another (marcianus) they are, for most of the length, on one only. In only two forms (angustirostris, sumichrasti) are they characteristically obsolete. The dorsal stripe, on the other hand, varies in width from one to five or six rows and is more often lacking. Neither the dorsal or laterals are absent individually except in melanistic specimens, being very constant in the forms which possess them. The head is usually of the same color above as the ground color of the back, but it is often more olive or brownish, and is occasionally marked with red or black. There are nearly always two proximate bright spots upon the parietal plates, although these are occasionally absent in individuals irrespective of race. On the sides of the head the dark color of the upper surface meets the usually much lighter color.of the superior labials, into which it grades on the preoculars, postoculars, and temporals. The labials are as a rule much lighter than above, the color being varying shades of green, yellow, blue, or occasionally white. They are nearly always more or less margined with darker, and there is a tendency, more pronounced in some forms than in others, for one of these dark margins to be extended upward along the anterior margin of the preoculars, and one along the pos- terior margin of the postoculars, making a conspicuous pattern on the sides of the head. 12 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. The ventral surface is ordinarily some shade of slate, often slightly iridescent, and frequently marked with red or black. The bases of the abdominal scutes nearly always possess a varying amount of black which is concealed by the overlying edge of the preceding scute. Where this pigment is well developed it forms a black bar along the base of each ventral scute, which is enlarged at either end into a spot that protrudes beyond the edge of the superincumbent scute, thus giving rise to the appearance of a row of spots along the sides of the belly. In some forms this bar is enlarged in the middle, and frequently to such an extent that it projects beyond the overlying scute and forms a median ventral band that varies in width from a narrow line to a broad band that covers most of the belly (elegans, melanogaster). The ground color of the belly blends with the color of the dorsal surface on the first row of scales. RANGE. The area inhabited by the genus Thamnophis, as at present known, includes all of North and Middle America north of the southern boundary of Guatemala and south of the fifty-second parallel of north latitude, although it is possible and even probable that both of these limits are surpassed. These snakes are known from almost every part of the intervening country except the lower part of the peninsula of Lower California® and the State of Sonora, Mexico, but the absence of records from both of these localities is probably due to the fact that these regions have been but little explored. The distribution of the different forms will be considered in detail when these are taken up. HABITS AND HABITAT RELATIONS. As in the case of most reptiles, very little is known of the habitat relations of the garter-snakes. Where detailed notes are available they will be given in the discussion of the different forms. In general it may be said that while the snakes of this group are not aquatic they apparently prefer the vicinity of water, as they occur in such places much more abundantly than elsewhere, although they are found not uncommonly on higher ground and in dryer situations. The extent to which they prefer damp situations also varies some- what with the form and the region, as will be shown later. Their food is in conformity with the nature of their habitat, and varies accordingly. Thus, individuals in the vicinity of water appar- ently subsist largely on frogs, tadpoles, and fish, while on the dryer uplands they have been observed to eat insects, earthworms, toads, a A single specimen has been recorded from Cape San Lucas, Lower California, but there is good reason to believe that there is an error in this record. (See p. 161.) VARIATIONS OF GARTER-SNAKES. 13 small mammals, and fledgling birds. While in captivity it seems to be impossible to get them voluntarily to eat dead food, but in the wild state specimens of three species (sirtalis, radix, elegans) have been observed to take frogs, small mammals, and birds that had been dead for a considerable time. In the northern part of their range, where cold winters prevail, they hibernate like other reptiles, usually passing the months of December, January, February, and part of March in a dormant con- dition. The actual time of hibernation is dependent upon the char- acter of the weather, for a few warm days in January or February is sufficient to bring them out. This was well illustrated in the winter of 1906, which was an exceptionally mild one, and garter-snakes were recorded in southern Michigan (Washtenaw County) on Jan- uary 22. At the approach of warm weather in the spring the garter- snakes are among the first snakes to appear, and may soon be found in numbers in the vicinity of marshes, ponds, and streams, feeding voraciously upon frogs that are so abundant in such places at this time of the year. But little is known regarding the time of coition, but at least one of the forms (sirtalis) is known to copulate in the spring, bringing forth its young in July, August, and September, and, as the other forms for which we have data also bear their broods at this time, the supposition seems justifiable that copulation in most cases takes place in the spring, notwithstanding the statement of Coues (1878, 278) that specimens of radix have been observed in coitu in Septem- ber and October. In some instances, at least, they exhibit a grega- rious tendency during the breeding season. In 1880, K. L. Ellicott (1880, 206-207) reported having seen an aggregation, in the follow- ing words: ‘‘I personally had the pleasure of observing it twice, both times very early in the spring, and in localities which could be called wilderness. I first saw such a bundle of snakes in the neighborhood of Ilchester, Howard County, Maryland, on the stony bank of the Pa- tapsco River, heaped together on arock and between bigstones. It was a very warm and sunny location, where a human being would scarcely disturb them. I reasoned that the warmth and silence of that secluded place brought them together. Some hundreds of them could be counted, and all of them I found in a very lively state of humor, hissing at me with threatening glances, with combined forces. and with such persistency that stones thrown upon them could not stop them nor alter the position of a single animal. They would make the proper movements and the stone would roll off; all the snakes in this lump were common snakes (HLutaenia sirtalis L.).”’ As Hay (1892 b, 528) remarks, ‘‘It is altogether probable that such assemblages are determined partly by the sexual impulses,” a view tliat is confirmed by the fact that in the spring groups do apparently 14 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. tend to be formed under the incitement of the sexual stimulus. In- deed Ellicott goes on to say, ‘‘The second time I noticed a ball of black snakes rolling slowly down a steep and stony hillside on the bank of the same river, but about 2 miles above Union Factory, Baltimore County, Maryland. Some of the snakes were of considerable length and thickness, and, as I noticed clearly, kept together by sbi impulses.”’ (The italics are mine.) The young are born alive, the usual number in each brood varying with the age of the snake, in the different forms and possibly in dif- ferent parts of the range of the same form. (See p. 191.) Dr. J. Schneck (1882, 1008) reeords a brood of 78 in sirtalis, and I have seen as few as four in butlerz. I have had opportunity of witnessing the birth of hundreds of young garter-snakes of several different species, and as the method was exactly the same in each case observed, there is little doubt but that it varies little throughout the genus. Just before parturition the female stretches out more or less, and may or may not crawl slowly about, but in either case maintains her tail curved slightly upward at the base, thus expanding the anal opening. Distinct labor periods occur, during which the young emerge singly, although a varying number may be extruded in a single fanaa In the majority of instances each young is still tightly toiled and in- vested in the fetal membrane when it appears, exactly as it has laid in the oviduct, although, owing to the occasional bursting of the sac during parturition, they sometimes appear to crawl from the anus. After birth the young snake lies quietly for a few seconds and then struggles lustily to thrust its head through the sac. If it is not suc- cessful it smothers (which occasionally occurs), but if it succeeds in breaking the sac it yawns once or twice, thrusts out its tongue, and crawls off, becoming at once very lively. As soon as the body be- comes dry, the skin is shed, exuviation often beginning within fifteen minutes after the young snake has been ushered into the world. The young snakes are apparently independent nearly from the first, probably, as Hay (1892 a, 389) has pointed out, owing to the ae that the yolk body is largely consumed before pas I have never noticed in the case of those born in captivity (parietalis, sirtalis, but- leri, radix, sackeni) a definite tendency in the young to remain near the mother for a time after birth, as has often been described. They almost immediately crawl away, hiding under sticks, stones, leaves, or other objects. Thus the occasional observations that have been recorded of a mother and her brood in the wild state were probably made immediately after the birth of the latter. After birth the young snakes are for some time very secretive, and show usually a very decided tendency to conceal themselves under objects, and even to burrow in loose dirt, when in confinement. This is without doubt the reason why specimens of about this age are so VARIATIONS OF GARTER-SNAKES. 15 rarely captured. As Abbott (1890, 296-297) says of these and some other snakes, ‘It is not a little curious that considering the large numbers of eggs that are found in the earlier part of summer so few small snakes are to be seen. I have never come across one less than 4 inches in length, except in the cases of the hog-nose snakes, men- tioned above, and of those which I have ushered into day by uncere- moniously breaking the egg before the occupants were quite ready to emerge. Even larger snakes, those measuring 6 to 10 inches in length, are by no means common.” Abbott is inclined to attribute this rarity of young snakes to enemies, but I believe it to be due, in large part at least, in the garter-snakes, to the retiring habits exhib- ited by the young snakes during this period of their life. In the case of those observed in captivity this secretiveness lasted until the first winter, and it is probably not until the following spring, at least, that they begin to seek their food in the open. In disposition the garter-snakes vary from the retiring, agile ribbon snakes to the more stocky aggressive species, such as sirtalis and radix. As a group, however, the adults are not retiring, a feature that combined with their abundance makes them the most common and widely known group of snakes in America. VARIATION. I have asserted that the first step in the search for relationships is to determine the variability of the characters. The traits that have been used in systematic work on the garter-snakes are the number of rows of dorsal scales, labial,ventral, and subcaudal scutes, the position of the stripes, comparative length of the chin shields, and proportionate length of the tail. It is now my purpose to subject each of these traits to an examination to determine whether or not they are variable, and, if so, the nature and extent of the modifications or variations. In investigating the variations it is essential that a distinction be made at the beginning between individual, sexual, and geographic variation, for it is evident that until such a distinction is made we can not hope to determine the value of the traits in pointing out affinities. In systematic mammalogy and ornithology these dis- tinctions are made, but unfortunately in snakes they have been largely ignored. For example, radix is generally described as possess- ing usually 7, but occasionally 8, supralabial scutes. Now it is con- ceivable that if no distinctions are made that the males might be characterized by 7 and the females by 8, or the opposite, that the variation is purely an individual one, or that there is an increase or decrease in the number in certain parts of the range, and until the actual conditions are ascertained there is little hope of determining affinities, for if in this particular form the variation is individual the 16 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. perpetual danger is encountered of correlating it with another in which the labials may also be said to be 7, occasionally 8, but in which this variation is geographic. As another example of this the usual formula given of the ventral scutes of different forms usually includes the extreme variation in number known in the group. It may be easily seen that this also offers no clue to affinities, for any geographic variation in the average that may exist and might serve to connect different forms is in this way completely obscured. The nature of the variations may be determined of course by taking the mean of the variation for the different sexes in different localities throughout the range, it being necessary to distinguish between the sexes, for if there is considerable sexual variation the predominance of either in the series from particular localities may obscure any geographic differences that may be present. This requires a large series of specimens from each locality, however, which we regret to say have rarely been available in this investigation. However, notwithstanding the fact that sufficient material has not been available in some instances to establish the actual extent of the range of variation of the sexes in the different localities, we believe that the data obtained is in nearly every case sufficient to indicate the nature and trend of the modifications exhibited by the different eroups. Variation in scutellation.—It is not surprising to find that it is largely upon the basis of the scutellation that most of the forms of Thamnophis have been defined, for the number of scale rows, labial, ventral, and subcaudal scutes have long been regarded by herpetolo- gists as available specific characters. Indeed, as will be brought out in the following discussion, these traits are, within limits, sufficiently stable to admit of their employment as diagnostic characters in the different forms of garter-snakes. Inasmuch, however, as similarity in the number of scales has been held to indicate genetic relation- ships, it is necessary to investigate the variations which they exhibit. The number of rows of dorsal scales has undoubtedly been con- sidered the most important systematic character in the determination of the different forms of this genus, as well as in many other groups of snakes. When an individual snake is examined these rows are found to be arranged in longitudinal series, running the entire length of the body, from the head to the tip of the tail. The scales of the different rows are alternately arranged, so that transverse as well as longitudinal series are formed. . Variation in number of dorsal scale rows.—The number of longi- tudinal scale rows on the body, i. e., anterior to the vent, is normally an odd number, as the series are usually paired on either side of a median dorsal row. If a specimen of some species (sirtalis, for exam- ple) be examined, it will be found that on the anterior half of the VARIATIONS OF GARTER-SNAKES. 7 body there are 19 rows of scales. If these rows be traced backward somewhat beyond the middle of the body, however, it will be noticed that a row is dropped on either side at approximately the same place, leaving 17. The decrease posteriorly in the number of scale rows, thus described for sirtalis, occurs in all of the garter-snakes, and the number usually given in diagnoses as typical of the different forms (23, 21, 19, or 17) is in every case the maximum number of rows, which, as we have just indicated, always occurs on the anterior half of the body. Furthermore, an examination of large series of these snakes has established beyond question that the decrease in the number of scale rows posteriorly is brought about in all of the forms of garter- snakes by the loss of certain definite rows. On all of the specimens examined with a maximum number of 23 scale rows anteriorly (megalops), which is the largest number that occurs in the genus, the fifth is first discontinued posteriorly, leaving 21, and shortly after- wards the sixth (now the fifth) leaving 19, and finally the fourth, leaving 17. In the specimens with 21 rows (elegans, radix, and hammondi, for example) it is the fifth row that is dropped first to leave 19, and then the fourth to leave 17. In those with 19 rows (sirtalis, eques, sauritus, etc.) the fourth is dropped posteriorly to leave 17, while when the maximum number of rows tends to be less than 19 (butleri, ordinoides, and scalaris) the fourth row drops out first to leave 17 and then the fifth (now the fourth) to leave 15. It will be seen at once that the order in which the rows are lost poste- riorly in the different forms is the same asin the form having the maxi- mum number of rows for the genus. In fig. 1 a diagram has been con- structed to show this law of reduction as it would be represented on a snake having a maximum number of 23 rows anteriorly and a minimum of 15 posteriorly—the extreme range of variation in the genus. An examination of this diagram shows that although the order in which the rows are lost-is, when counted from the ventral series each time, as is usually done, 5, 5, 4, 4, the real sequence when expressed in terms of the maximum number of rows for the genus is 5,6, 4,7. The diagram also illustrates the manner in which the rows are lost, for in the great majority of cases the scales of the row which is dropped become smaller and finally cease, the adjacent rows con- verging to occupy the space. At the point where the row is discon- tinued there is often, however, a rather large scale bearing two keels, indicating that the last scale of the lost row has fused with its neighbor. In case such a fusion occurs it may be either with the row above or below, but the fact that the scales of the row to be dropped usually become smaller toward the termination of the series, and that the keel of this row is usually only on the edge of the large scale, generally denotes clearly which row is really lost. 18 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. The decrease in the number of scale rows posteriorly is, we believe, Fic. 1.—DIAGRAM ILLUSTRATING THE ARRANGEMENT OF THE DORSAL SCALE ROWS IN THE GARTER-SNAKES. posteriorly, thus making a formula THE FIGURE MAY BE TAKEN AS REPRESENTING A SECTION OF THE TRUNK, EITHER ANTERIOR OR POSTERIOR FROM THE MIDDLE OF THE BODY, OF A HYPOTHETICAL INDIVIDUAL WITH THE MAXIMUM AND MINIMUM NUMBER OF ROWS FOR THE GENUS. correlated with the taper of the body, for as the scales themselves decrease but little in size toward the tail the loss of the rows is a necessary re- sult of the shortening of the metameres. Furthermore, as shown above, the reduc- tion posteriorly takes place caudad to the middle of the body, 1. e., where the body tapers to the tail. This phe- nomenon may not in itself be considered important, but, as will be shown directly, it has a wider significance than as an explanation of the man- . ner in which the scutellation accommodates itself to the posterior diminution in the size of the body. In certain forms the maxi- mum number of scale rows is known to be very constant. For example, specimens of sirtalis, eques, proximus, sau- ritus, or sackeni are but very rarely found that have either more or less than 19 rows an- teriorly, but, notwithstand- ing the fact that most of the other forms are also usually described as having a definite number of scale rows, there is considerable variation in this respect. If a specimen of megalops with a maximum number of 23 rows be exam- ined it will be found that this — number only occurs on the middle of the body, owing to the fact that the fifth row is dropped anteriorly as well as of 21-23-21-19-17, counting the VARIATIONS OF GARTER-SNAKES. 19 maximum number of rows on the different parts of the body.¢ Similarly also in variable forms that have a maximum number of 21 rows, the fifth is often dropped anteriorly as well as pos- teriorly, giving a formula of 19-21-19-17. In the forms having a maximum of 19 rows the formula may be 17—19-17-15, with the fourth row shortened anteriorly as well as posteriorly, while in those in which 17 is the maximum it is also the fourth that is dropped anteriorly when there are but 15 anteriorly. The abridged row when shortened anteriorly varies in length, and may extend nearly to the head or be abbreviated so as to include but a few scales on the middle of the body. This suggests at once that among the specimens in a given locality those in which a certain row is dropped anteriorly as well as posteriorly represent an intermediate condition between those in which this row is continued to the head and those in which the maximum number is one row on either side less than the maximum for the locality, the decrease being brought about by the loss of the first abridged row in the maximum formula. To illustrate, individuals of radix from the vicinity of St. Louis, Mis- sourl, may have the formula 21-19-17, 19-21-19-17, or 19-17, the length of the fifth row varying in the specimens with 19-21-19-17 rows from a few millimeters to include the entire anterior half of the body, and we can only conclude that the formula 21-19-17 is a result of the continuation of this row to the head, while in specimens with 19-17 rows it is entirely lacking. It is thus the first abridged row that is shortened and lost in the decrease in the number of scale rows, and as this row, as we have . previously seen, generally ends a little beyond the middle of the body, it is evident that most of the shortening must take place at the anterior end of the row. There is, however, a posterior shortening of the second abridged row, for upon the disappearance of the first short row the second abridged series becomes the first, and its pos- terior end must move up toward the middle of the body to retain the symmetry. This variation in the maximum number of scale rows may be summarized as follows: The decrease in the number of rows of dorsal scales among the garter-snakes of a given locality is due to the anterior shortening and subsequent loss of the first row of scales that is dropped posteriorly in each snake having the next higher scale formula. I have been unable to discover any sexual variation in the number of scale rows, but in many cases an examination of large series reveal distinct geographic differences in this regard. Thus in the great plains region radix exhibits the scale formulas 21-19-17 or 19-21-19-17, but in the prairie region to the east a decidedly larger proportion of specimens have 19-21-19-17 rows, while near ae eastern limits of its range the formula 19-17 is | occasionally a4In the plows paper hee rows will be Pempercd from ihe gastrosteges each time. 20 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. encountered. More than this on the eastern boundary the species is replaced by a smaller form (butlert) which is not known to have a higher formula than 19-17. I will not take the space here to give further illustrations, as they will develop in the consideration of the various forms. It is sufficient to say that the geographical variation in the number of scale rows, shown to occur in radix, also occurs in many other forms in the genus. Since the geographic variation in the number of scale rows is brought about in the same manner for the various forms as in the individuals of a given locality, and it further develops that these forms at points in their range where the number of scale rows is at a minimum or maximum often grade into, or are replaced by, other evidently nearly related forms with a smaller or larger number of rows, the conclusion seems to follow that the racial differences in the number of dorsal scale rows in the garter-snakes are the result of the loss of the abridged rows in the order in which they are dropped posteriorly in individuals that have the maximum number of rows for the genus. At first sight it appears rather odd that the first abridged row should end so constantly just beyond the middle of the body, and that any further decrease should take place from the fore backward, but an explanation presents itself. As has been previously stated, the reduction in the number of rows posteriorly is correlated with the taper of the body. If, therefore, we consider that there is also a slight decrease in the size of the body anteriorly that culminates in the constriction of the neck, the tendency for the rows to drop out anteriorly may be considered analogous to their abridgment posteriorly. This anterior taper is slight and not sufficient in many specimens to make a difference in the number of rows between the neck and body, in which case the first abridged row extends to the head, but it is apparently such that a decrease in size would effect the scale rows here before on the middle of the body. There is no noticeable difference in the size of the scales between individuals with a greater or fewer number of scale rows, but it is very noticeable that those forms which possess the smallest formula in the genus are distinctly the smaller in size, and the op- posite. It seems safe to conclude, therefore, that the loss of the rows is correlated with a general decrease in the size of the body, and that in the decrease the number of rows is influenced by the necessity of maintaining the symmetry of the body, so that the decrease is noticeable first on the anterior and posterior parts of the body, owing to the comparatively smaller size of these parts. Whether or not such dwarfing takes place we will be more able to judge when the evidence of the other characters and the forms themselves have been examined. VARIATIONS OF GARTER-SNAKHES. 21 However, since, as I shall show later, the males are smaller, more elongate and less stocky in form than the females, a tendency toward a reduction in the number of scale rows should, it seems to me, be revealed first in the former sex. This point is difficult to test, for extensive series are wanting in most regions where such a reduc- tion is inaugurated, and in a region where the reduction is great enough to markedly influence the females the difference between the sexes would be obscured, since there is not enough difference in relative size to give the males the next lower formula. A series of specimens of radix from Palo Alto and Clay counties, lowa, however (where, as will be shown later, there is a slight tendency toward a reduction in the number of scale rows from 21-19-17 to 19-21-19-17), seem to confirm our reasoning, for out of seventeen males thirteen have the formula 19-21-19-17, as against 21-19-17 in the others, while out of thirty-five females but six have this formula, the re- mainder’ having 21-19-17. This matter should receive further study. The law of variation to which the scale rows are subject may be formulated as follows: The indurdual, geographic, and racial varia- tions in the number of dorsal scale rows in the garter-snakes is brought about by the shortening and loss of the same scale rows as are ordinarily dropped posteriorly in conformity with the taper of the body, and there is evidence that this decrease is due to a dwarfing of the body. It should be pointed out that the adjustment of the scale rows to the taper of the. body shown in these snakes is very similar to the plate arrangement observed in certain Paleechinoids by Jackson and Jagger (1896). If our diagram (fig. 1) be compared with their diagram of the ideal arrangement of interambulacral plates in Melonites multiporus (p. 164) it will be seen that the rows dropped posteriorly in the snakes are almost exactly in the middle (of each side), as the rows of inter- ambulacral plates which are discontinued toward the ventral end in the echinoids are in the middle of the interambulacral area, so that the order in which they are dropped in the snakes and echini is exactly the same. In the echinoid figured, owing to the pronounced con- striction dorsally, the columns of plates are distorted, but continue to be represented—a feature that is not shown in those garter-snakes in which the maximum number of rows continues to the head, possibly because the constriction of the neck is too slight to disturb the rows. That a dying out of columns also occurs in echinoids, however, is stated by Jackson and Jagger, and it is very interesting to note that ‘““when columns die out or cease to be continued to the dorsal area it is commonly the middle or last added column which drops out first in the cases observed,” which is exactly analogous to the dropping of the scale rows anteriorly in the garter-snakes, 83553—Bull. 61—08——3 22 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. Variation in number of labial plates.—The number of labial plates (fig. 2) in the garter-snakes varies from 6 to 9 above the mouth and 8 to 11 (very rarely 12) below, each of the numerous forms tending to have a certain definite number, as in the case of the scale rows. For the latter reason the number in each: of these series is often used as a diagnostic character, but, owing to the fact that individual variations are not uncommon, this trait has been accorded only secondary impor- tance, while of the two the number of supralabials is considered of more taxonomic importance, as it is noticeably less variable. There seems to have been little attempt, however, to determine the nature of these variations. ; In almost. any locality a series of specimens @ will reveal differences in the number of these plates. Thus, in proximus, while the labial formula ° in the great majority of specimens from the same region is Prencular Gecondrnw 9/20, there maybe 7 Ras tral Poe taei inet End ee 5; lower labials. In Prenasal | yee ean auritus the formula ] emporal temporals Seum Postnasall Temp is mostly 7/10, but specimens with 8 su- perior and 9 or 11 inferior labials occa- sionally occur. Again, ordinoides ] e Sn rae specimens from the UG Ot ae nee ers same locality may labials Tnfra- lab ials have 6, 7, or 8 supe- FG. 2.—THE ARRANGEMENT OF THE HEAD PLATES IN THE GENUS THAM- TOL and 8, 9, or 10 NOPHIS, AS REPRESENTED IN AN INDIVIDUAL POSSESSING THE MAX- jnferjorlabials. The IMUM NUMBER OF LABIAL PLATES FOR THE GENUS. variability shown by these forms is typical of all of the forms in the genus, and it will be seen at once that the amount of individual variation is very slight. In the case of the lower labials the difference indicated in the speci- mens from the same locality may consist of both a loss and an addi- tion of a scute to the normal number for the locality, but it rarely exceeds this amount. In the case of the supralabials, however, the difference shown by specimens from the same locality usually consists of either an increase or decrease of one scute, it being rarely that the variation occurs both above and below the normal number for the region. This apparent definiteness in the variation in the number of upper labials at once suggests the presence of some controlling factor. It «The following discussion may be followed on the diagram that forms fig. 3. 6 The labial formula is the number of scutes in each series, and is expressed as a fraction; for example, 8/10 means 8 supralabials and 10 infralabials. VARIATIONS OF GARTER-SNAKES, 23 can not be sex, for in no instance where large series have been exam- ined has it been possible to detect any correlation between the number of labials and the sex of the specimens. On the other hand, when the racial differences are examined the apparent definite- ness in the individual variation of the upper labials is seen to be extended also to the lower. Thus, in proxvmus, which as a form has 8, occa- Formula sionally 7, supralabials, and 10, but also 9 or 11, infralabials, the presence of 9 lower labials occurs much less fre- quently than 11 throughout most of the range, while in sauritus, on the other hand, in which the upper labials are 7, occasionally 8, and the lower 10, occasionally 9 or 11, specimens with 9 infralabials are much more abundant 8-10 than those with 11. Sauritus thus exhibits a tendency toward a smaller ees number of lower as well as upper labial 8-9 scutes than proximus. Furthermore, where the ranges of these forms meet and the forms intergrade (in Illinois, for example), the specimens appear to be about equally divided in the posses- sion of 7 or 8 supralabials, while the infralabials are normally 10, with the occasional variations 9 and 11 about equally disposed. This at once sug- gests that these variations are geo- graphic. It is difficult to demonstrate a geo- eraphic variation in the number of labial scutes within the different forms, for the amount of the individual dif- ferences is usually equal to any geo- ¢.g eraphic variations that exist in the Sammentorn, \Phisonecessitates «the vex-- "1% DSR SHOWING THE MERHOD amination of large series of specimens siat scuTEs THAT RESULTS IN THE tobtain as nearas possible the mean FO ovrrouus ov aaavunanacns, number for each locality, for in small series of specimens slight fluctuations in the locality averages, due to the accidental preponderance of specimens with the same number, are sufficient to obscure any geographic trend in the variations. In cer- tain intergrading forms, however, it is possible to show that geographic differences do occur within the form. There is good reason to believe, - \ ‘ ! 1) [eelsistel7tetst Sl" lia sé [7 Bator ¢ Be. eR el ole hel BN 32 Ze al oN | [Gd | oie <0 K, 2 N Zo] =o i sai a om eG =|2) N“N als oO a= Bn) ao sls an Ze zie Z Sel. = Museum), sHowinG THE RE- . . DUCED THIRD SUPRALABIAL—THE FIRST mally ‘7 and occasionally 'S (rade), OF» \nakcn nc sam urpoceion peo eTOre usually 7 and frequently 8 (ordinovdes), be scrutinized, it will often be noticed that in the specimens with 8 the third is frequently decidedly narrower than the second or fourth (fig. 6). Both in front and behind the eye in such specimens the supralabials are arranged exactly as in those which have normally 8, except that the common suture of the second and third occurs near the posterior margin of the loreal, thus narrowing the third plate very decidedly. In these forms =< when there are but 7 scutes there is ORvEnun’ or MicineawMossumy, either no labial suture below the loreal, SHOWING THE NORMAL ARRANGEMENT the second plate extending from the FORMULA 1379. Posterior ‘nasal to the preocular, or the posterior margin of the first labial is moved slightly backward so as to lie under the extreme anterior part of the loreal (fig. 7). There can be no doubt, therefore, that the decrease in the number of supralabials from 8 to 7 is in these cases due to a loss of the third scute and that. the method of this loss consists principally of a narrowing of this \ i | wee fa) = eg a ge ae ie wma PIES VARIATIONS OF GARTER-SNAKES. 27 plate. That fusion also plays a part is also very evident, how- ever, as illustrated by fig. 8, but that this usually takes place only when the plate in question has be- come much reduced in width also appears evident, for specimens have i, not been observed in which either of PRS : ea) (ES GAS Say Bay a the adjacent plates were strikingly : ~ Fig. 8..—THAMNOPHIS ORDINOIDES (1109 enlarged, as must have been the Case FIELD MUSEUM), SHOWING THIRD SUPRA- nad: this plate fused with ‘either ofits: 142” Bepucky AND PARTY FUSED With THE FOURTH TO LEAVE SEVEN. neighbors before becoming reduced. The fact that the adjacent plates are not noticeably larger in speci- mens with 7 than in those with 8 supralabials also makes it very probable that with this loss of a supralabial in front of the eye there is associated a shortening of the muz- zle. In forms which have normally 7 the arrangement is exactly the same as Same 7-—"__ in specimens with 7 in the forms which Fia. aa epee oe UNL have Tor8 (fig. 9). VERSITY OF MICHIGAN MusEUM), SHOW- b ae ING THE NORMAL ARRANGEMENT OF THE There is but one form (butlerz) in oe es EN Ae REMUS Sy nieh ate supralabials are very fre- quently 6, but radix (and apparently also ordinoides) occasionally exhibits a tendency to have less than 7. If a specimen of radix or butleri (fig. 10) with 6 labials be examined, it will be found that the first four scutes are arranged exactly as in specimens with 7 (fig. 7). The fifth plate, however, is somewhat wider than usual, extending to beyond the middle of the first temporal instead of in front of it, while the next, which is now the last, ee extends to the middle of the third as usual... io taawwormis BUT: In butleri, and frequently in radix, when there — verr (226 RutHveN Cor are 7 labials the same arrangement of the {icon wer soenechmar scutes is adhered to, except that the fifth in causep sy rae pscrEase these individuals extends from the postocular Yen curs anp irs rosion to a point in front of the middle of the first 9 wr me anrerenutt temporal (its normal position in forms with 7 © “*** “4*!4% scutes in this series), while the next extends from here to beyond the middle of this scute, the last retaining the position which it has in specimens with 6 labials (fig. 11). It would seem from this that the decrease in the number of labials from 7 to 6 has resulted from {/ PD um SSe 4@In many specimens of butleri the last labial ends below the single large second temporal, but this scute has very evidently been formed by a fusion of the scales which usually constitute the first and second rows behind the first temporal, so that the relative position of the posterior margin of the last supralabial in respect to these scales is the same as in other forms which do not have a single temporal in the second row. 28 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. the fusion of the penultimate and antepenultimate scutes. While such a fusion undoubtedly exists; however, it constitutes but a part of the story; for, if simply a fusion of these scutes took place, the fifth, in individuals with 6 labials, should extend from the postocular to the second temporal, while, as it is, it only extends a little beyond the middle of the first temporal, being thus but slightly larger than in specimens with 7. A reduction in the size of the penultimate scute must therefore take place at some time during the process of reduc- tion in the number from 7 to 6. A series of specimens of butleri and iia a eS Fig. 11.—THAMNOPHIS RADIX (155 RUTHVEN COLLECTION), SHOWING SIXTH SUPRALABIAL NARROWER THAN USUAL AND ENDING ON THE FIRST INSTEAD OF THE SECOND TEMPORAL. radix from the common boundary of their range, where, as we have seen, a reduction in the number of labial plates from 7 to 6-takes place, will illustrate this process in all of its stages. Thus, in most of the specimens with 7, the penultimate scute extends from the anterior part of the first temporal to the second row of temporals, the normal arrangement for this number of labials (fig. 9). In other specimens (fig. 11) this labial extends only from the anterior part of the first temporal to beyond the middle of this scute, while in still others the reduced penultimate scute is partially fused with the antepenultimate one (fig. 12). As the posterior margin of the penultimate plate, in specimens with 6 labials, generally comes in contact with the first temporal at about the place where the posterior margin of the sixth occurs in the specimens in which it is much reduced, the next step in the reduction of this plate would seem to be its entire fusion with the preceding scute to leave 6 (fig. 10).¢ Fie. 12—THamnornis Burt- Therefore the reduction in the number of labial LERI, SHOWING THE LAST STAGE IN THE LOSS OF THE scutes from 7 to6 wn the garter-snakes vs brought — sixtH supRaLaBiAL—THE about by the narrowing and loss of the sixth; and FUSION OF THE REDUCED SCUTE WITH THE FIFTH. as there is not a corresponding increase in the ; size of the neighboring scutes, this decrease in the number of labials must be accompanied “by a shortening of the mouth behind the eye. From its apparent general occurrence among the different forms, the above described method of variation in the number of supra- labials may be formulated in the following general terms: A reduc- tion in the number of supralabials takes place in the garter-snakes as a od a1 have previously noted (Ruthven, 1904, 294) that when there are 7 supralabials in butleri, the extra one is apparently formed by the division of the fifth. The expla- nation for this is now shown to lie in the method of reduction in the number of these scutes from 7 to 6. VARIATIONS OF GARTER-SNAKES. 29 result of the narrowing and final loss of the fourth, third, and eighth scutes consecutively, as numbered on a snake with 9 plates in this serves. I have previously noted that the number of labials in the upper and lower series is correlated, and that a reduction in the supra- labials is associated with a decrease in the number of infralabials. The method of reduction in the number of the inferior labials is, how- ever, somewhat more obscure, for, owing to the small size of these plates, it is often difficult to detect any decrease in the size of indi- vidual scutes, so that in forms with a different number of scutes in the lower series it is very difficult to say just which plates have been lost. Much light can be thrown upon the problem, however, by cor- relating the position of the individual plates of the two series.¢ If a specimen of megalops with the labial formula 9/11 be examined it may be noted that the upper margin of each individual scute in the lower series corresponds almost exactly in width and position to one in the upper series, except that the first and seventh supralabials are above two scutes in the lower series. This arrangement is quite accurately maintained in all of the specimens with this formula exam- ined. If, however, a specimen with the labial formula 9/10 be scru- tinized it will be found that, while the same arrangement is for the most part adhered to, the small fourth labial has not in these speci- mens a corresponding infralabial, but that the preceding infralabial scute (the fourth) extends backward to the posterior margin of the fourth supralabial (fig. 5). Evidently, then, either the third or the fourth infralabials are not present in these specimens, or they have fused into a single scute. Just which method prevails is imma- terial, for we hope to show that in other cases, as well as in this one, the reduction in infralabials always takes place beneath the upper labial scute that is being reduced; and as was seen in the case of the upper labials that the upper scute is first reduced and finally fused, it may be taken for granted that it is the corresponding infralabial that is lost. In specimens with the labial formula 8/10 the third superior has a single corresponding inferior labial scute (fig. 4), and there is in such specimens a condition exactly similar to 9/11 forms, except that the fourth supralabial and its corresponding infralabial (the fifth) are lost. In those rather rare cases where the formula 8/11 is shown, it is inter- esting to note that there are two infralabials below the third supra- labial, so that the arrangement is exactly similar in 9/11 forms except that the fourth supralabial only has disappeared. In the specimens that show a tendency toward a reduction in the labial formula below 8 B/ 10 the third superior, as already noted, becomes aThe ee been: out in fine iemee discussion are tiger ate oh in the diagram (fig. 3). 30 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. noticeably smaller. In such specimens when the formula is 8/9, the - reduced third superior scute (as in the case of the reduced fourth in the formula 9/10) hasnot a corresponding inferior labial, but is included within the limits of the preceding inferior scute, which also underlies the second supralabial. Here again is a loss of an inferior labial scute corresponding in position to a reduced superior labial. As already seen, it is the reduced third supralabial that is lost to make 7 upper labials, and in the specimens with the formula 7/9 the disposition of the inferior plates in front of the eye is exactly as in the formula 8/10, except that the inferior labial corresponding to the third supralabial in the 8/10 specimens is lacking (fig.7), and likewise the same as in 9/11 specimens, except for the loss of the fourth superior and its corresponding inferior scute. In specimens with the formula 7/10 (fig. 9), a fourth infralabial is present, but with the third lies entirely under the second supralabial, while in 7/11 a third, fourth, and fifth may be seen beneath the second supralabial. Up to this point the reduction in the number of inferior labials, as in the case of the upper labials, has taken place in front of the eye. When the labial formula is 7/9, however, and the penultimate superior labial is narrowed, there is likewise a reduction in the width of the penultimate scute in the inferior series. When the small penultimate supralabial has disappeared, leaving 6, an infralabial scute also tends to disappear, leaving 8, and the same arrangement as in 7/9 specimens is attained, 1. e., each superior labial except two are situated above an infralabial (fig. 10); but in this instance it is the first and penulti- mate supralabials which have two corresponding inferior scutes, instead of the first and antepenultimate, as in other instances. In this instance, however, it frequently does not appear that the lost inferior labial is the one corresponding in position to the lost supra- labial, for in many 7/9 specimens in which the penultimate supra- labial is reduced in size the penultimate inferior labial extends partly under the antepenultimate superior scute, thus occupying the place of the posterior small inferior labial that usually underlies this scute and with which it has evidently fused. The penultimate infralabial is, however, first reduced in size at the same time that the corresponding supralabial is pushed forward in the process of reduction, so that it may be considered as the one that is lost in the reduction in the number of infralabials from 9 to 8. The reduction in the number of lower labial scutes in the garter- snakes thus takes place in exactly the same manner and at about the same time and place as the decrease in the number of superior lahials, which explains the existing correlation in the number of plates vn the two serves. Examining the diagram (fig. 3) as a whole, it may be seen (1) that the reduction in the two series takes place at about the same time and place, and in the following order from the maximum number: VARIATIONS OF GARTER-SNAKES. Sih 4/5, 3/4, 8/10; (2) that the loss of the scutes in both series takes place first in front and then behind the eye, which retains the same position in regard to the superior and inferior plates; (3) that as the reduction consists of a decrease in the size, and final loss, of the disappearing scutes, without a corresponding increase in the size of the remaining scutes, a contraction of the mouth is necessarily associated with the decrease in the number of labials. We can even go farther than this, however, for in these snakes the posterior angle of the mouth extends practically to the posterior angle of the jaw, which in turn marks the extreme posterior limit of the head. This is shown externally by the fact that the cephalic plates never extend beyond the posterior angle of the mouth. Since the number of labials is correlated with the length of the mouth, which extends the entire length of the head, it logically follows that the number of labials is also an expression of the head length. It is very difficult to prove the existence of any geographic or racial differences in the length of the mouth or head by actual measure- ments, as these differences must in most instances be very slight, as indicated by the limited range of the variations in the number of labials. It is apparent, however, that such differences exist, for two forms, butleri and ordinoides, are characterized by the noticeably small size of these features, and it is to be noted that besides being the only forms in the genus with a decidedly small mouth and head, these forms also possess the smallest number of labials, and if a comparison be made between these rather extreme forms and their nearest rela- tives, radix and elegans, a difference in the length of the head and mouth may be readily seen. ‘ From the above discussion of the method of variation in the number of labials the law of labial reduction for the genus may be formulated as follows: The differences in the number of labial plates in the garter- snakes is the result of the loss of corresponding superior and inferior latial scutes, the loss taking place both vn front and behind the eye, thus resulting in a shortening of the mouth and head toward this organ. Variation in number of ventral and subcaudal plates and in propor- tionate tail-length.—As: previously stated, there is a single series of large scutes on the belly and two rows of smaller ones on the ventral surface of the tail. The number of scutes in each of these series has been considered of systematic importance, and it is true that each form has a definite number within limits. The range of individual and sexual variation is so great, however, and the racial differences gen- erally so slight that large series must be examined from each locality to determine geographic variation. The number of ventrals is apparently less variable than the sub- caudals. There is some sexual difference, as may be determined by an examination of a series of specimens from any locality, but the 32 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. limits of this variation overlap considerably owing to the individual - variation. Thus, in a large series of radixfrom Clay and Palo Alto counties, Lowa, the ventrals in the females range from 148-166, in the males from 160-172. In order to determine whether or not geo- graphic variation occurs in a species it Is necessary to note the pro- portion of sexes in the locality averages, since a predominence of either obscures slight geographic differences. It can be shown, how- ever, that geographic variation not only exists but that it is corre- lated with the reduction in the scutellation shown above. Concrete examples will be brought out in considering the different groups. It is only necessary to point out here that all of the forms which have a much reduced number of dorsal scales and labial plates have also a proportionately small number of ventral plates, and the opposite. On account of their greater variability, therefore, the number of scutes in this series is a trait which is less useful to the student than the number of dorsal scale rows and labial scutes, and, while by no means valueless, is to be used principally to substantiate the evidence of the other characters, when allowance has been made for the dis- turbing factors of individual and sexual variations. The variation in the number of subcaudals is of the same nature as that of the ventrals in that the number is greater, as a rule, in males, the extremes of variation in the two sexes overlap, and the individual variation is considerable. In the case of the urosteges, however, the number is not always correlated with the reduction in the scutella- tion, but with the tail length, which may vary to some extent irrespec- tive of the body. The most pronounced case of this is in the inter- grading varieties of the ribbon-snake group. Sauritus in Michigan has a proportionately longer tail and more subcaudals than the south- western form proximus, which has more ventrals and labials and attains a much larger size. The character has thus always to be checked up with measurements, but it is not valueless, for when the same proportionate length of tail ismaintained the number of urosteges is correlated with the rest of the scutellation. Happily the propor- tion of tail to body length is quite constant throughout the genus; the only exception being the ribbon-snakes, which usually have a tail length of .27-.38 of the total length and are sharply separated by this and the large number of urosteges from the rest of the species in the genus, which have a tail length of .19-.27, the females of the former approaching the males in the latter. Variation in number of preoculars and in arrangement of lateral spots.—The number of preoculars has been employed but little in specific diagnoses, although it has been used frequently in descriptions. The explanation for this is to be found in the fact previously stated that in most of the species in the genus there is but a single plate in front of the eye. Certain forms, however, (me- —<* VARIATIONS OF GARTER-SNAKES. oo lanogaster, angustirostris, hammondi, ordinoides, and elegans) either quite constantly or in some part of their range, exhibit more than one preocular. Whatever may be the cause of this division of the pre- ocular, other evidence, as will be shown later, appears to indicate plainly that these fons are closely related, so that the presence of two or more preoculars over all or a soi of the range separates them from all others in the genus. Although the arrangement of the lateral spots are usually referred to in descriptions, the character is entirely too variable to be used extensively in diagnoses. As before stated, when distinct they are in the form of two rows of alternating spots on either side between the stripes. In forms where they are ordinarily distinct, however, they may be fused to a variable extent in melanic individuals, while being more than usually prominent in light-colored ones. While they are thus quite variable individually, they also exhibit some racial differ- ences. Thus, in parietalis the upper row is usually fused into a band, and in eques, sirtalis, and parietalis the first few anteriorly frequently tend to fuse with their neighbors in the opposite row to form cross- bars. In two forms the arrangement has become modified into a series of transverse blotches that usually in one form (scalaris) stop at the dorsal stripe but in another (phenax) crosses the back and fuses with its fellow on the opposite side, forming crossbars. As in the case of the number of preoculars, not enough is known about the factors which influence the arrangement of these spots to enable us to pin our faith to them as indicative of racial affinities; as also with the number of preoculars, however, the fusion of the Sites into transverse blotches for the entire length occurs in forms which other evidence seems to indicate are algsely related, so that the presence of this arrangement seems to group these forms in a natural way. Furthermore, I believe (my reasons are given on pp. 120-158) that the forms that have this arrangement of the lateral spots are directly related to those that tend to have more than a single pre- ocular, in which case the two characters when used together serve to define a natural section of the genus. Variation in stripes.—Ever since the establishment of the genus Eutaena by Baird and Girard, the possession of three longitudinal light color bands has been considered a distinctive feature of the genus and the particularities of these stripes used as diagnostic characters. They are, as we have seen, disposed as a single dorsal and two lateral bands, the dorsal covering the median and usually more or less of the adjacent rows, when present, the laterals a varying amount of the second, third, and fourth rows on either side. The peculiarities of the dorsal stripe that have been used in sys- tematic work are its presence or absence and width, and these char- 34 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. acters are fairly constant in the different forms. For example, in marcianus, ordinoides, and elegans it is usually narrow, while in radix and sirtalis it is decidedly wider, and while it is present in most eroups, it is almost universally absent in hammondi, melanogaster, and sumichrasti. It is thus often of considerable valte in distinguish- ing the different forms that occupy the same region. [Even in its more general aspects, however, it is unreliable as a test of affinities, as is attested by the considerable range of variation in width which it frequently exhibits in the same form. Indeed, in some instances it tends to be nearly or entirely lost in certain parts of the range of a form in which it is elsewhere distinct, so that an attempt to‘use it as a test of genetic relationships must result in the grouping together of widely divergent forms. The causes of these variations in the dorsal stripe are difficult to trace without a knowledge of habitats. In certain melanistic forms, such as concinnus, it is present, but narrow. On.the other hand, in certain quite pallid types (elegans and eques) it is also quite narrow. As a rule, however, in most melanic forms it is either narrow, obscure, or absent, although it is also obsolete in at least one form in the mountains of Mexico (swmi- chrasti) which is decidedly not melanistic. It is hardly possible to plot on a map the regions where the dorsal stripe tends to be lost, for the tendency varies with the form, while in the case of melanism the cause is apparently not always a climatic one. For example, in northeastern United States the genus tends to become decidedly darker colored. In sirtalis specimens this darkening of the general color is associated with the narrowing and frequent loss. of the dorsal stripe, but no such tendency is exhibited by sauritus, which occupies the same region. Similarily, also, although swm- chrasti on Mount Orizaba has lost its dorsal stripe, it is distinct in scalaris specimens from the same region. The forms of the Mexican plateau all usually show the dorsal stripe distinctly with the excep- tion of melanogaster, in which it is characteristically lacking. In this form, however, it accompanies a melanistic coloration which can not be due to climate, as this is a desert region, but is possibly asso- ciated with the more aquatic habitat of this form. In this connec- tion it is interesting to note that the dorsal stripe in megalops, in which the dorsal is generally very well defined, is usually very much obscured in specimens taken from Lakes Chalco and Patzcuaro, Mexico, where the form tends to be much darker, and also seems to be quite aquatic in its habits. The conclusion seems justified, there- fore, that the dorsal stripe is probably modified by several factors, one of these being the general darkening of the body color. At any rate, the same peculiarities arise independently in different forms not directly related, thus invalidating these characters as indications of affinities, VARIATIONS OF GARTER-SNAKHS. 35 As in the case of the dorsal stripe, the position of the lateral color bands has been made the basis of specific diagnoses, for it was appar- ently recognized quite early that the position of these stripes was con- stant in the different forms. Thus, Baird and Girard (1853) noted that in the ribbon-snakes its position was on the third and fourth rows, in distinction from the other forms then known, in which it was upon the second and third rows;* while Cope, Brown, Stejneger, and others have continued to make use of it as a distinctive character as the genus became enlarged. The position of this stripe has been reported by Taylor (1892, 320) to be subject to occasional variations (apparently meaning here in- dividual variation), but this conclusion may probably be taken as a result of an attempt to group specimens of apparently similar but in reality distinct forms, for there can be no doubt of the really remark- able fixity of this trait. In not a single specimen in the 3,000 that have been examined has the position of the lateral stripe varied as much as a single row from the normal for the form.° There seems to be some evidence that the position of this stripe is affected by the factors that influence the dorsal, for in northeastern, northwestern, and southwestern United States, and the mountains of eastern Mexico there is a tendency toward a narrowing and loss of the lateral stripe in certain forms, that closely parallels similar conditions in the dorsal band. This is not other than is to be expected, from the fact that the stripes very probably have the same physiological significance, but, be this as it may, the extent of the modifications is so decidedly less in the case of the lateral stripe that there are but two forms (angustirostris and phenax) known at the present time in which it is characteristically absent, none in which it is on other than the second, third, and fourth rows, two (marcianus, butleri) in which it is difficult to say whether it is upon the second and third or third and fourth rows, and but one (buwtleri) of the latter in which there can be much question as to which of the two positions it is to be referred. The position of the lateral stripe is thus apparently a very good and reliable speciiic character, owing to its aus within the aBaird a Gira erre Ei in iain ae (the Ae Aaa listed in ine: @ Waloue. of North American Reptiles besides the ribbon-snakes that has the stripe on the third and fourth rows) with those having the bands on the second and third rows. b Hay (1892, b522) records a specimen of sauritus from Wabash County, Indiana, which had the stripe on the fourth and fifth rows. It should be noted, however, that this snake had also 21 scale rows, which is such an unusual number among the ribbon- snakes that in the large numbers examined we have never observed a specimen with over 19. It is thus very probable that the individual was abnormal as regards the arrangement of the dorsal scale rows, and the possibility is at once suggested that an extra row may have been added on either side below the lateral stripe, which would be a decided abnormality. 36 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. limits of the different forms. It may still be subject to modifications, however, that cause it to vary between the different forms. Should such changes occur they may be conceived to take place in two ways— either independently or dependently of the scale rows upon which they are situated. If they are independent of particular scale rows, changes in their position might take place either suddenly or gradu- ally. A gradual change ought to show forms that are characterized by a different arrangement of the stripe intergrading with each other in this regard, unless, indeed, they have since become distinct. An examination of the intergrading forms, however, fails to reveal any- thing of this nature. Thus, sirtalis-parietalis-concinnus, elegans- ordinoides, and proximus-sauritus can be shown to intergrade along their common boundaries, and yet there is no change in the position of the lateral stripes, the intergradents being characterized by the same arrangement of these bands. Similarly, also, in the case of the forms now apparently isolated (megalops and hammond, for example, see pp. 123 and 137) the evidence of the other characters and of geo- graphic probability point much more strongly toward a relationship with the forms in which the lateral stripes have the same position than with others. These evidences of relationship also argue against the possibility of a sudden change in the position of the stripes as regards the scale rows, for, as far as I can see, there is no evidence of rela- tionship between any two forms with the bands on different rows that is strong enough to require us to adopt this view for which we have no direct evidence. This apparent constancy in the position of the lateral stripes both within the different forms and between re- lated forms gives strong support to the view that these bands are closely associated with particular scale rows. Even if the lateral stripes are associated with certain scale rows, however, there is still the possibility that their position may be changed by the loss or addition of certain of these rows, but a mo- ments consideration will show that the chance for any great modifi- cation in this way is slight. At present, as we have seen, the fourth is the lowest row to be dropped in the reduction in the number of scale rows. The position of the stripe when on the second and third rows can, therefore, experience no change in a decrease in the number of scale rows from 23 to 15, the range of variation in the genus, for it is only the fourth, fifth, sixth, and seventh that are lost im this reduction. In those forms in which the stripes are upon the third and fourth there is also little chance for modification, for in but one instance is the fourth row of scales dropped anteriorly to leave 17. It is, however, in this group that most of the modifications in the position of these stripes occur, and these variations are associated VARIATIONS OF GARTER-SNAKES. ai with the reduction in the number of scale rows. Thus, where the fourth row is dropped posteriorly to leave 17 the lateral stripe in one form (radia) is left upon the third rew only, in two forms (butlert and megalops) it descends below the third to include the second, while in three forms (proximus, sauritus, and sackenz) it ascends one row after the loss of the fourth, so as to remain posteriorly also upon the third and fourth. In but two forms is the position of the stripes modified anteriorly, and in one of these (marcianus) it is apparently due to the general pallidness of the ground color which partially obscures it ; in the other (butlert) we have the only decidedly dwarfed representa- tive of the forms which have the stripe upon the third and fourth rows. When the seale formula in butleri 1s 19-17, the lateral stripe is ante- riorly upon the third and the margins of the second and fourth rows, posteriorly upon the second and third. When the scale formula is decreased to 17-19-17, the lateral bands are anteriorly and posteriorly upon the second and third rows, on the middle of the body being still upon the second, third, and fourth. In the forms that have the lat- eral stripes upon the third and fourth rows there is thus apparently exhibited a tendency toward producing a form with the stripe upon the second and third rows, but as this only takes place in the most dwarfed forms and is evidently due to dwarfing it can in no way be taken as lessening the gap between the groups that have it upon the second and third, and those in which it is upon the third and fourth, for it would be absurd to maintain that the former were dwarfed de- scendants of the latter, since they are by no means all dwarfed forms. I have shown that the other traits (number of scales, color, and tail length) are subject to modifying influences, and although of assistance are not to be depended upon as indicating related groups. If, however, I have succeded in establishing that the position of the lateral stripe is not affected by the various influences that divide the genus into many and diverse forms, I should find in it a trait that will when availiable give us a true idea of the general relationships of the different forms. @ Variation in color.—There is considerable variation in the speci- mens examined, but this is so largely individual that it is difficult to define a normal color for a form, and, even when there are well marked geographic differences among the forms, those in the same nO 4 Coues and Yarrow (1878, p. 273) were apparently the first to point out the signifi- cance of the position of the lateral stripe, although they did not pretend that the members of the different groups were genetically related. The following is their state- ment of the case: “As far as we have seen, the position of the lateral stripe may be a means of grouping the-species. Though this varies within certain limits, mainly according to the width of the band, yet its position on the third and fourth, or on the second and third, dorsal rows affords a ready means of distinguishing certain sets of species or varieties.” The validity of this statement has been questioned by Taylor (1892, 320), but, as indicated above, his criticism was probably not well founded. 33553—Bull. 61—08——4 38 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. region tend to be similarly colored, as Allen has pointed out a number of times in mammals and birds, so that it is impossible to distinguish them sharply on this basis. The following unquestionable tend- encies may be noted: . A marked increase in bright colors in the Pacific coast region in ee and Oregon, intl British Columbia. This is shown by the presence of bright reds and greens (concinnus, ordinoides). An increase in the amount of black pigment at the expense of the paler colors in the region just mentioned. This is especially shown in the enlargement of the lateral spots to the obliteration of the interspaces on the skin and a narrowing of the dorsal stripe (con- cinnus, ordinoides). 3. A tendency toward a paler ground color and lighter stripes in the forms inhabiting western Texas, southern New Mexico, southern Arizona, and northern Mexico (marcianus, eques). 4. A tendency toward the production of red pigment on the Great Plains (parietalis and occasionally radix). 5. A tendency toward dark colors in the forest region of eastern United States. This is shown by the prevailing brown ground color in sauritus and butleri, and the dark greenish and brownish olives in sirtalis. An increased darkness of color thus occurs in the more humid areas, while the paler colors are usually found in the more arid regions; some well marked exceptions occur, however. Thus, in the arid region of northern Mexico angustirostris and melanogaster, which are very dark in color, occur with the paler megalops and eques, and in central Texas prorimus is much darker than marcianus. More detailed notes on these forms are needed before much can be done by way of explanation, but it is significant that the dark forms (melanogaster, angustirostris, and proximus) mentioned above as occurring with pale ones in an arid region are known to be more than usually aquatic in habits. We believe that it will be found that the color in these snakes is not, as in mammals and birds, closely correlated with the major environmental complexes, owing to the fact that they prefer damp situations, and are thus not exposed to the same conditions throughout their range as terrestrial mammals and birds. At any rate, owing to the great variability, color is one of the least important characters that have been used in diagnoses. As is well known, the characters that have been discussed are by no means all that have been used by students of the group to define species. They are, however, those which we have found to be of the most importance. The number of temporals and comparative length of fore and hind chin shields, while both of use in defining certain species, are too variable or the peculiarities of too limited occurrence to permit of their use in a search for general relationships. The VARIATIONS OF GARTER-SNAKES. 39 shape of the frontal (i. e., length in relation to width) is also useless * as a diagnostic character, owing to its variability; while, as shown above, the position of the eye relative to the supralabials (i. e., the particular labials which enter the orbit), is dependent upon the number of labials in this series. THE FOUR GROUPS OF GARTER-SNAKES. It may seem the extreme of ‘“lumping,”’ in view of the thirty forms’ that are at present recognized by herpetologists, to assert that there are but four great groups or lines of descent in the garter- snakes, but I believe that the evidence is sufficient to warrant the assertion. The first step in the division of the genus into genetic groups is to determine the different associations of traits, or forms, that exist at the present time. This has of course already been largely accomplished in the definition of the species and varieties that have been made, and it remains for us now to assemble these forms into the general groups.? The table that follows expresses all of the different combinations of characters that are shown by these snakes at the present time, and it will be seen that there are nineteen of these combinations or forms rec- ognized. These comprise all of the different forms of garter-snakes that have a geographic range, and disposes of the multitude of forms that have been described upon the basis of individual or sexual variations. @ This number has been attained by adding the better defined Mexican and Central American forms to the North American forms admitted by Brown (1901) in his “Review of the Genera and Species of American Snakes North of Mexico”’ (a con- servative estimate ). bIt is best at the outstart to ignore all questions of species and subspecies until their status is established, and to speak of these as forms. Forms, therefore, in the sense employed in this paper, are actual combinations of traits, having geographic extent, irrespective of whether they are isolated (species) or intergrade with their neighbors (subspecies). Detailed discussions of questions of nomenclature are also omitted, although the names used are in every case the ones that, in the light of the develop- ment of these investigations, we judge to be the right ones, following the Interna- tional Code of Zoological Nomenclature. The proper name of each form will be found in the footnotes, together with the synonymy. BULLETIN 61, UNITED STATES NATIONAL MUSEUM. 40 LI-61-1G-6T | | | UpuoMUDy | | | LI-61-16 ——— | | | 8-(6) Ler-e-61 SLLISOLUYISNOUD LI-61-1Z | 2T=61=10-60-16 snyinvs 1-(g) — LI-61 "4000 Jed JZ SNYLANDS' Wayans (1)=8 uey} o10Uur a Ay] B10Ue*) J -) snuaxo.d (L)-8 igh gr a LI-61 jen) -- — a —s —_ I oy S quaying o-2 ST-LT-61-L1 : LT-61 a LI-61-L1 21-61 ee ae LT-6I-1G-61 “quad Jed 1z LT-61-16 UBY} SSo] xypoy ane lew rs Ay[B10u9+) SNUDPILDUL 8 LI-61-1Z LI-61 sdojpbaum 8-(6) eae LI-61-16-£6-16 *soqnos ; *SUOISIAI A snues ened *sdnoiy *SuIIO.T He ress aeons eae *[184 Jo Y43u9e] o1vu0IOdo1g JequinN *SIBQ-SSO10 SB posuelie JOU sjods [8103 P| pues auo SIB[NIOII J "SMOI YJ INOJ pue pryy aq) WO *syods [B10}®] JO JUOMIASUBIIB PUB S1e[NOOIIg ‘ad114s [B10] B] JO UOIPISOg "SayDUS-lajDb ay} Ur suorsiaip pun ‘sdnoub ‘swiof o7Ur S74} JO WOLWDUIGQULOD AY} 7D1]SN]]0 0} 2790 L, 4] VARIATIONS OF GARTER-SNAKES. THAMNOPHIS II syDIIs sunday | STSZT USDLYIVUNS 8 LTSOLeeE CI-LI-61 snuujouoo *SIBQ-SSD10 SB or : *qu00 Jed JZ posuviie jou ue} sso sqyods [¥1018 sypjayuvd | 1-(8) LI-6I gee eS ad Ay[B10te+) pus | 9uo sIB[nd0e1 S2yDpL28 | Ono 1-8 CI-LI-6I-L1 LI-61 CT-LT : CI-LL-6E-21 SILD]DIS 1-(8) 11-61 LI-61-1¢-61 apuayd 1-8 LI-61 1aj}sD60UD)a UL 8 CT-LI-ST ‘asUBI OY cT-LT yo 41vd owos saprourp.o 9-1-8 CT-LI-GI-LI *4yue0 rod 12z UI ‘s1Bq-Ssso10 : 11-61 uByy ssol SB posuviie LT-61-1G-61. Ayye1000 5 sj0ds [81030] B 10 11-61 ‘OM4 SIV[NIO0I I supb9)9 (1)-8 LI-6I-1Z-61 LT-O1-16 “SMOI PITY pus puodes oyu} UO 42 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. In the discussion of the value of the characters an attempt was made to show that the position of the lateral stripe groups related forms, and that the proportionate tail-length, and the number of preoculars and presence or absence of lateral cross-bars,” are also of value in this regard, if used within the limits of the groups having the lateral stripe in the same position. If we have made our point, then, the nineteen forms defined in this table fall naturally into four groups, the members of which should be genetically related. To each of these four groups of garter-snakes we have for conveni- ence given the name of the best known form which it includes. Thus we have the Radix group with the lateral stripe upon the third and fourth scale rows and a short tail; the Sauritus group with the lateral stripe on the third and fourth rows and with a long tail; the Elegans group with the lateral stripe upon the second and third rows, and with either more than one preocular scute in some part of the range or the lateral spots arranged as cross-bars in some part of the range, and the Sirtalis group with the lateral stripe on the second and third rows, and never more than a single preocular, or lateral cross-bars. It will probably be conceded at once that the ribbon-snake group (Sauritus) is a natural one, and there can be little question that the forms included in the Radix group are also related, although the placing of marcianus with these snakes will be seriously questioned by some herpetologists. In the case of the other groups, Llegans and Sirtalis, the affinities are in many cases much less obvious, but it can not be denied that as thus defined the groups include those forms which are evidently related. Thus there is no question but that hammondi, ordinoides, and elegans, which are here placed in the Elegans group, are closely related, nor that sirtalis, parietalis, and concinnus, in the Sirtalis group, actually intergrade. The evidence for the inclusion of the other forms in these groups, as well as further proof that the above-mentioned forms are properly placed together, will be given in more detail when the forms are discussed separately, and the lines of interrelationship considered. The continuous geographic range of the four groups also seems to strengthen the evidence of the characters that they are associations of directly related forms. As is shown on the maps, the Radix group occurs on the Mexican plateau, great plains, and in the prairie region; Sauritus is a coastal plains group, extending from British Honduras to Maine, only penetrating into the interior in the forests of eastern United States and along the streams of the great plains; the Alegans group occupies the Mexican plateau and, in the United States, all of a It has been a source of regret to the writer that a single trait could not be found to separate the Hlegans and Sirtalis groups. Most students of snakes have encountered a similar difficulty, however, for the different groups of snakes, apparently owing to the recency of the order, are frequently not sharply defined by any single difference that is obvious. VARIATIONS OF GARTER-SNAKES. 43 the region west of the Rocky Mountains, only occurring east of these mountains on the high plateaus; the Sirtalis group occupies the plateau region in Guatemala and Mexico, and in addition practically all of the United States and southern Canada, from the Atlantic to the Pacific. The range of each of these groups, as will be shown later, is made up of the ranges of the component forms placed side by side. The evidence seems to be quite conclusive, therefore, that these groups are assemblages of directly related forms, and it is necessary now to consider in detail the forms comprising these groups in order to determine the directions of their relationships. If this detailed examination bears out the evidence that the groups are genetic, and reveals the direction of the affinities within the groups, much light should be thrown upon the origin and history of these snakes. A key is given by which the forms may be recognized. Key to the forms of the genus Thamnophis. a‘. Lateral stripe anteriorly upon the third and fourth rows. b'. Tail generally more than .27 of the total length. c!. Supralabials usually 8. d'. Tail between .25 and .35 of the total length............... proximus, p. 98. (Ilinois to western Nebraska, southward in eastern Mexico to Central America. ) d:. Tail between .32 and .38 of the total length. ............ .sackeni, p. 107. (Principally confined to Florida.) e. Supralabials usually 7......-. ae aes aan ohare Se tay Ci ee sauritus, p. 112. (United States east of the 87th parallel, and north of Florida.) b’. Tail generally less than .27 of the total length. cl. Dorsal scale rows usually 21-19-17, supralabials 8, occasionally 9... . - megalops, p. 44. (Southern Arizona and New Mexico, the plateau of Mexico.) ce. Dorsal scale rows usually 19-21-19-17, supralabials 7 or 8. .......radix, p. 70. (Great plains and prairie regions of central North America.) c®. Dorsal scale rows usually 19-17, supralabials 6 or 7. ......-..-- -butleri, p. 87. (Indiana, Ohio, southern Michigan, western Pennsylvania. ) a’. Lateral stripe anteriorly upon the third row only ...........---- marcianus, p. 58. (Central Texas, southern New Mexico and Arizona, northern Mexico.) a’. Lateral stripe upon the second and third rows. b'. Dorsal scale rows usually 21-19-17 or 19-21-19-17. c'. Dorsal stripe usually absent. d'. Kye in contact with two supralabials................... hammondi, p. 133. (Southern California and northern Lower California.) d?. Eye in contact with a single supralabial............. angustirostris, p. 120. (Northern part of the Mexican plateau, southern New Mexico and Arizona.) e+ Dorsal Semperusially PRCSeCMG-. =). oot. eee - secs cee eee es elegans, p. 138. (From thé Rocky Mountains to the Sierra Nevada-Cascade Range, inclusive.) b?. Dorsal scale rows usually 19-17 or less. ‘ c'. Dorsal stripe usually present. d'. Never more than 1 preocular. CAceupialapiaisnisually 82-286. so .2 cess cafe teke------. /. .eques, p. 158. (Guatemala to southern Arizona and New Mexico on the plateaus.) ’ e?, Supralabials usually 7. 44 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. f'. Lateral spots fused into transverse blotches for the entire length. . scalaris, p. 127. . (Southern Mexico and northern Central America. ) jf’. Lateral spots not fused into transverse blotches for the entire length. g'. Both rows of lateral spots distinct on the skin; interspaces not gen- erally fed... 22.2242. 22 . S20 ek Bigs ee ee ea Oe so aalinise (North America east of 91st meridian and south of the 52nd par- allel.) g°?. Upper row of lateral spots usually fused on the skin; interspaces red. h'. Dorsal stripe usually covering one and two half-rows. Width of interspaces on skin equal to the length of one scale .....- parietalis, p. 166. (From the 91st meridian to the Pacific coast, exclusive of western Oregon and Washington, and south of the 52nd par- allel. ) h? Dorsal stripe usually confined to one row. Width of interspaces on skin generally less than the length of one scale, frequently ab- SOME» (ee tac See ieee Pa tae eee ee ee nae concinnus, p. 173. (Western Oregon and Washington. ) d? Frequently 2: preoculars.. 2-25 .¥2s 22 see ees ee ordinoides, p. 147. (California, Oregon, and Washington, west of the Sierra Nevada-Cascade Range. ) c* Dorsal stripe usually absent. d‘ Dorsal scale rows 19-17. e! Lateral spots fused into transverse bars, which usually extend entirely across thesbacke:*.Ae....22. os oniskack Saat ee So ae phenax, p. 130. (Veracruz, Mexico.) e? Lateral spots not fused into transverse bars ......-.--- melanogaster, p. 124. (Southern Mexico.) d? Dorsal scale rows usually 17-19-17 or 17-15......--..._-: sumichrasti, p. 164, (Southern Mexico.) THE RADIX GROUP (MEGALOPS, MARCIANUS, RADIX, AND BUTLERI). MEG ALOPS.« Description.—The Mexican plateau, possibly on account of its topo- graphic unity, is the home of one of the most homogeneous forms in the genus, notwithstanding the fact that this form has been divided into four species (megalops and macrostemma Kennicott, and flavi- labris and insigniarum Cope). This form, if we succeed in establish- ing its unity, should be known as megalops, and is undoubtedly the largest of the garter-snakes. It is characterized in general by having the lateral stripe on the third and fourth rows, mostly 21-19-17 scale rows, 8 or 9 supralabials, 152 to 174 ventrals, 60 to 87 subecaudals, and a proportionate tail-length of .20 to .278. There is considerable variation in color, so that no single description can be given that will apply for the form. For detailed description see page 55. Habits and habitat relations.—As is usually the case among reptiles, but little has been recorded concerning the habits of megalops, which a Thamnophis megalops (KrNnnicott), Proc. Acad. Nat. Sci. Phila., 1860, p. 330. In- cludes E. macrostemma Kennicort, E. insigniarum Corr, and E£. flavilabris Corr. VARIATIONS OF GARTER-SNAKES. 45 is unfortunate, as a knowledge of its habitat would undoubtedly throw much light upon the explanation of its distribution. At first sight it seems rather surprising that a species of this genus should occur so abundantly in an arid region, and three alternative explana- tions present themselves. It may be either a distinctive desert type and confined to arid localities even in the more humid (southern) portions of its range, or it may, on the other hand, be addicted to a more aquatic habitat, being confined in the desert regions to the vicinity of water. As a third alternative it may be that the form is able to adjust itself to either arid or moist conditions, a change that if extensive might necessitate a change of habits. The second explanation is probably most nearly the true one, for the genus is notably inclined toward a moist habitat. Cope in 1885 (1885a, 386-387), writing of the habits of this snake at Lake Xochi- milco, near the City of Mexico, says, ‘The ends and shores of the piers are the resting places of innumerable water snakes, which can readily be observed froma canoe. * * * We caught a considerable number and found that they belong to the two species above named [insignia- rum and melanogaster]. The EF. insigniarum is the most active, sooner seeking the water, where it swims, keeping close to the shore, and remaining more or less in sight until it conceals itself in a hole. * * * The food of both of these species is the Rana montezuma Baird and another species allied to Rana halecina.”’ Again (1900, 1030) he writes: “On being disturbed the EF. macro- stemma plunges into the water, but does not go far beneath the sur- face, but takes refuge under the edge of a bank, or emerges in a new spot, so that it is difficult of capture. The columns that support the aqueduct that carries water from Chapultepec to the City of Mexico are covered with a dense vegetation, which is continuously watered by leaks in the venerable structure. On examining this vegetation at my height above the ground, I encountered in the thick of it a round eye. [Exposure revealed first the head and then the body of a snake of this species, which found a congenial abode there.”’ Baker in 1895 (1895, 120-121), in speaking of a trip on Lake Patz- cuaro says, ‘‘As we approached the island, a commotion was ob- served in the water, and a snake was seen to glide swiftly and noise- lessly away. I quickly shot it; on picking it up we found it to be a large species of water snake (/. insigniarum). As we pushed our boat among the reeds bordering the island a great commotion was created in the water and a number of these snakes were seen to swim away.” From these accounts and the large numbers of specimens in different museums from the lakes of southern Mexico, there can be no doubt that megalops in southern Mexico not only thrives in a moist locality but is also quite aquatic in its habits. In contrast with these descrip- 46 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. tions is that of Doctor Coues (1875, 613), who states that according to his observations the species in Arizona is not “specially addicted to the water, frequenting dry herbage and bushes.”’ This statement may justly be withheld as evidence, however, for, as will be shown later, the identity of the species which Coues referred to is doubtful. Since Cope’s Duck Creek, New Mexico, specimen was also found in swampy ground in the vicinity of water, we may conclude that as far as our present knowledge goes megalops throughout its range prefers the vicinity of water. If this is true, while of more general distribution in the southern parts of its range (States of Mexico, Puebla, ete.), to the northward it probably becomes confined to those habitats in which water is to be found for the greater part of the year. Range.—Much.of the unity of this form may be accounted for by the character of the region which it inhabits. Geographically the Mexican plateau is a tableland lying at an elevation of 3,000 to 8,000 feet, and separated from the Gulf of Mexico on the east, and the Gulf of California and the Pacific Ocean on the west, by parallel mountain chains and a narrow strip of coastal plain. To the north it descends gently into the open basins of southern Arizona and southwestern New Mexico, being limited in its northward extension by the escarpment of the Colorado plateau, thus including the Proplateau region of southern United States (see Ruthven, 1907). Southward it extends to the isthmus of Tehauntepec, where it ends by the junction of the two parietal mountain ranges. This plateau has been built of the accumulated waste of the preexisting mountains, and volcanic discharges, the existing mountain peaks protruding through this mass, mainly in the higher parts of the plateau, as nunataks in an ice field (Heilprin, 1902, 774-775). The climate of the plateau may be characterized in general as tem- perate (Heilprin, 1900, 777). The amount of precipitation is vari- able. In southern Mexico it is comparatively great, the rainy season lasting about six months, and the streams and lakes are perpetual. To the northward, however, owing to the influence of the eastern chain of the Sierra Madres, much of the moisture carried by the warm gulf winds is precipitated on the coastal plain, so that in their subsequent passage over the tableland they are desiccating winds. During the short rainy season of three months the flora of this arid region grows luxuriantly, but during the long dry season, in which the region is exposed to the glare of the tropical sun and the hot, dry winds, the vegetation becomes parched, and the streams and lakes partially or entirely dry. The aridity of this area increases to the northward, from San Luis Potosi northward being known as the Chihuahuan desert that grades into the Proplateau region of south- western United States. This entire region, therefore (northern part of the Mexican plateau and the Proplateau regions), is arid. The VARIATIONS OF GARTER-SNAKES. 47 plains and lower mountain slopes are the home of the cactus, mes- ‘quite, yucca; creosote bush, and prickly pear, while the summits and canyons support a luxuriant growth of hardwoods or pines, according to the altitude. Specimens have been examined from Tucson and Yuma, Arizona; Duck Creek, New Mexico; Lerdo, Durango and Coyotes, Durango; Guanajuato; Ocatlan, Jalisco; Lake Chalco, Chapultepec, City of Mexico, Chalco, Lake Xochilmilco, and Lerma, Mexico; Patzcuaro, Michoacan; Puebla, Puebla; ‘‘South Mountain or Micrado,’’ Vera- cruz; Colonia Juarez, Jimenez, Chihuahua, San Andreas, and Minaco, Chihuahua. From these records the range of megalops may apparently be defined as the entire southern part of the Mexican plateau, extending northward along the western part of the plateau into southern Arizona and New Mexico. Owing to the fact that no real boundary of the tableland exists on the north, no geographical line of demarcation can be drawn here, and we find the species ranging into the open basins of southern Arizona and southwestern New Mexico, as far west as Yuma and to the north possibly to the escarp- ment of the Colorado plateau (the northernmost records being Fort Whipple, Arizona, and Duck Creek, New Mexico). Coues (1875, 615) recorded the species from Fort Whipple, Arizona, and Cope deter- mined his specimens as this form. We are unable either to find these specimens or any others from this locality, so that, in view of the difficulty that has existed in properly distinguishing the garter- snakes of the southwest, this record should be held in abeyance until supported or denied by further evidence. Whether or not the form extends entirely across the northern part of the Mexican plateau, i. e., into the States of Coahuila and San Luis Potosi, remains to be discovered. As we stated in the consideration of the environmental conditions, there are no geographic barriers in an east and west direction on the plateau, and megalops probably ranges over the entire area (Fig. 13). Variation.—The uniformity of topography is clearly expressed in the homogeneity of scutellation, and the variation when properly determined elucidates many of the questions that have arisen over the division of the form. Bearing in mind that the males have as a rule a longer tail and more ventral and subcaudal plates than the females, and making allowance for the irregularities caused by a marked dominance of either in the averages, it will become evident that ‘a great similarity in scutellation prevails all over the plateau from Puebla to Arizona. The scale rows, only in rare instances, vary from 21-19-17 by the presence or absence of a row; the ventrals vary from 152-173 with an average of about 160-165; the tail length is about .23 of the total length in the males and .21 in the females, and the number of labials oscillates about an average of 8 or 8.5. So 48 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. closely is this unity preserved over the entire extent of the range that it is only by carefully plotting the data from large series that any definite trend in the variations can be observed. When this is done, however, there is seen to be a distinct although slight decrease in scutellation and tail length toward the south. iy hs a Be Ai “1 is Ke “ey tite) Ye * ses on eh ae x :: Erase ——— TAC : ao ee ‘ . ons Route a se = Sz poe Ss EF s_.J ‘\ —_— = =n y = aor St FIG. 13.—DISTRIBUTION OF THAMNOPHIS MEGALOPS, AS INDICATED BY THE LOCALITY RECORDS. From Arizona to Patzcuaro, Michoacan, no specimen has been examined with less than 21-19-17 scale rows, but 6 out of 45 speci- mens from the intervening region exhibit a tendency to have more than this number, in the presence of an extra row on either side for a varying distance, thus making the formula 21—23-21-19-17, the extra row never extending to the head. South of this locality but VARIATIONS OF GARTER-SNAKES. 49 one specimen has been examined with more than 21-19-17 rows, while about the City of Mexico and in Puebla a considerable propor- tion have the reduced formula of 19—-21-19-17, and in three specimens from Veracruz one has 21-19-17, one 19-21-19-17, and the third 19-17, the smallest number for the species. It is also of interest to note here that the type of flavilabris which was from Veracruz (State) was said to have had 19 scale rows (Cope, 1866, 306). In the diagram that constitutes fig. 14 we have attempted to represent eraphically this geographical variation in the number of scale rows. The number of specimens at our disposal may perhaps seem too small upon which to base conclusions as to the variations in the ~ number of scale rows, since it has often been affirmed that the number may vary a row or two in the same locality. Bearing in mind, how- ever, the narrow limits of individual variation shown in our locality AN a) aes 9 [Sea Bile lie he pile al se Wd al7/ Tueson, Colonia Coyotes, Patzcu- Vicinity Puebla, Veracruz Ariz. Juarez, Durango; aro, of Puebla. (State). San An- Durango. Michoa- Mexico dreas, can. City. Minaca, Chihua- hua; Chihua- hua. Fic. 14.—DIAGRAM SHOWING THE VARIATION IN THE DORSAL SCALE FORMULA IN THAMNOPHIS : MEGALOPS. records and indicated on our diagram, it may be clearly seen, even from this limited series, that, although slight, the decrease in the number of rows from Patzcuaro southward is entirely definite; while a similar reduction is indicated to the northward that may or may. not be confirmed by larger series. In the number of labials, as in the case of the dorsal rows, the question of sex may be ignored and the averages plotted directly, as has been done on the diagram, fig. 15. By comparing the tables it will be seen that although the arithmetical mean is quite constant, it is not uncommon to find 9 supralabials and 10 infralabials in Chihuahua; that no specimen north of Patzcuaro has been observed with less than 8, and that south of this locality 9 labials very seldom 50 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. occur. This variation is so slight that were it not for the considerable number of specimens upon which the averages have been based it might almost be ignored. However, it indicates, we believe, that there is a geographical variation in these plates, correlated with the decrease in the number of scale rows to the southward in Mexico. As 9 Tucson, Colonia Ju- Coyotes, Patzcuaro, Vicinity of Puebla, Ariz. arez, Durango; Michoacan. Mexico City. Puebla. San Andreas, Durango. Minaca, Chihuahua; Chihuahua. Fic. 15.— DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUPRALABIALS IN THAMNOPHIS MEGALOPS. we have before stated, the number of infralabial plates is more subject to individual variation and requires a larger series than is here avail- able to furnish the mean number for each locality. Fig. 16 will show the amount of variation in the material examined. Tucson, Colonia Ju- Coyotes, Patzcuaro, Vicinity of Puebla, Ariz. arez, Durango; Michoacan. Mexico City. Puebla. San Andreas, Durango. Minaca, Chihuahua; Chihuahua. Fig. 16-—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF INFRALABIALS IN THAMNOPHIS MEGALOPS. In regard to the ventral scutes a similar variation is to be observed, but, as indicated above, the question of the sex of the specimens examined must be carefully noted. In general it may be said that there are in the males from 0 to 17 more plates than in the females, VARIATIONS OF GARTER-SNAKES. 51 so that if one sex greatly predominates in a locality it may throw the average either above or below that of another locality in which the opposite sex prevails, and thus obscure any geographical variation that may exist unless very pronounced. Even when each sex is plotted by itself geographical variation may be obscured by the individual varia- tion owing to the small number of specimens. If, however, it can be shown that there is a similar geographic variation in both sexes and this cause of fluctuation is removed by combining the two sexes in approximately equal proportions, the increased amount of material should reduce the individual variation and also tend to decrease both the sexual and individual variations below the geographic. Colonia Juarez, Coyotes, Patzcuaro, Vicinity of Mex- Puebla, San Andreas, Durango; Michoacan. ico City. Puebla. Minaca, Durango. Chihuahua; Chihuahua. Fic. 17.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF VENTRAL SCUTES IN THAMNOPHIS MEGALOPS. When the range of variation in the two sexes is examined separately it is found that in Chihuahua the average number of gastrosteges in the males is 170 and in the females 165, a condition that seems to prevail southward nearly to Patzcuaro; at the City of Mexico, how- ever, it is 165 in the males and 160 in the females, which indicates a reduction in gastrosteges in both sexes very similar to that of the scale rows. In diagram, fig. 17, I have plotted the locality variations in the number of these plates, and, although the sexes are only approxi- mately equally represented in each series, I believe that the mean represents nearly the average number for each locality. If I am right in this it will be seen that there isa gradual decrease in the number of ventrals from Chihuahua and Patzcuaro toward Mexico City and - 52 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. Puebla. While a similar reduction is indicated in my material from southern Arizona, the series upon which the computations are based is too small to be relied upon as representative. The variation in the number of subcaudal plates is very similar to that of the ventral plates, but, owing to the much wider range of both sexual and individual variations, the geographic is even more liable to be obscured. In general the males may be said to have from none to about 10 urosteges more than the females, while the individual variation in the males may be as high as 12 and in the females 8. 72 Colonia Juarez, Coyotes, Patzcuaro, Vicinity of Puebla, San Andreas, Durango; Michoacan, Mexico City. Puebla. Minaca, Durango. Chihuahua; Chihuahua. FIG. 18.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUBCAUDAL SCUTES IN THAM- NOPHIS MEGALOPS. If we take only those localities from which seven or more specimens have been examined and examine the range of variation in the two sexes, there is shown that in Chihuahua and Durango the variation is from 75 to 87 (males) and 66 to 74 (females), while at the City of Mexico and Puebla it is 67 to 79 (males) and 61 to 67 (females), thus seeming to reveal a slight geographic variation. If males and females be combined in approximately equal numbers (fig. 18), an average of 76.3 is given for Chihuahua and 66.6 for Mexico City, which shows a shght decrease in harmony with the reduction in VARIATIONS OF GARTER-SNAKES, 53 scutellation, observed in the case of the scale rows, labials, and gastrosteges. To summarize, then, there is exhibited in this form a decided tendency toward a decrease in the number of scale rows toward the southern part of the Mexican plateau, which is correlated with an apparent decrease in the number of labial, ventral, and subcaudal plates, and the locality records suggest that a similar reduction in the number of scale rows and ventrals exist in southern Arizona. In general I think that it can be safely affirmed that this reduction in scutellation to the southward is an evidence of dwarfingin this form, although it is impossible to prove it, of course, by measurements. The proportionate tail length of specimens from the various localities has been examined and seems to be quite constant through- out the entire province, although an apparent geographic variation in this trait seems to be demonstrable. In Chihuahua it varies from .24 to .27 of the total length in the males, and in the females from .20 to .23, while about the cities of Mexico and Puebla it drops -.26 Colonia Juarez, Coyotes, Patzcuaro, Vicinity of Puebla, San Andreas, Durango; Michoacan, Mexico City, Puebla, Minaca, Durango. Chihuahua ; Chihuahua. Fic. 19.— DIAGRAM SHOWING THE VARIATION IN THE PROPORTIONATE TAIL LENGTH IN THAMNO- PHIS MEGALOPS. to .20 to .22 (females) and .21 to-.24 (males). When these are averaged in equal numbers and plotted (fig. 19), a shght decrease in the average length of the tail to the southward is evident. It is true that this reduction only amounts to .03, but it must be noted that the sexual and the individual variations have been excluded in large part, while a reduction approximately equal to the extremes of variation in either sex in one locality is still indicated. This decrease in the length of tail should not be confused with the de- crease in the number of dorsal rows, labials, or gastrosteges, of which it is entirely independent, but it possibly explains in part the reduction in the number of subcaudals. When the scutellation and tail length is plotted, the value*of the divisions that have been made of the group on the basis of these characters is at once revealed. In 1860 Kennicott (1860, 330) described a specimen labeled ‘‘Tucson and Santa Magde”’ (No. 965, U.S.N.M.), collected by Major Emory and A, Schott, which had a 33553—Bull. 61—08-——6 54 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. proportionate tail length of .25, giving it the name of megalops. Immediately ‘lllseatns this he described a specimen from the City of Mexico, by Major Ries giving no measurements, probably owing to the fact that the tail was aprokem In 1885 Pibiosser Cope (1885b, 173), in giving a synopsis of these forms, listed as megalops a speci- men collected by himself on Duck Creek, New Mexico, which had a tail length of .26, and stated that macrostemma (insigniarum) may be distinguished by its shorter tail. In 1886 Cope (1886, 285) again listed as megalops ten specimens collected by Wilkinson at Chihuahua and stated that they did not differ from the Duck Creek specimen. In 1892 (1892, 646), in a key to the species of garter-snakes, he gives as the distinguishing differences that megalops has a tail length of less than one-third and more than one-fourth the total length, while in macrostemma it is more than one-fourth and not less than one- fifth, and on a subsequent page (651) lists three specimens of macro- stemma (insigniarum) said to be from near Prescott, Arizona, thus extending the range of the latter to cover most of the region occupied by the former. In 1901 Brown (1901, 22), in endeavoring to straighten out the American representatives of the group, found in five specimens from Mexico “‘ the tail to be about one-fourth of the length, or longer than in most adult megalops, which reverses the proportions given by Cope.” This threw the Arizona (insigniarum) specimens of Cope, which as Brown:shows are from Tucson instead of Prescott, with megalops, which now becomes distinguished from macrostemma by the presence of a shorter tail; macrostemma is excluded from the United States, being replaced by megalops in southern New Mexico and Arizona. The solution of this tangle is not hard to find if we take into con- sideration the sex of the specimens. In the first place, both Ken- nicott’s type of megalops and Cope’s Duck Creek specimens are males, the latter having a slightly longer tail than is usual, while the type of macrostemma (although not measured), and a number of other specimens from southern Mexico examined by Cope were females, while at least some of those that he measured (type of insigniarum) from the latter region, although males, have short tails, owing to the shortening that we have shown to take place ea the southern part of fe range. Finally, Brown’s Tucson specimens were females with characteristically short tails, which led him to reverse Cope’s description. As a matter of fact there is, as shown by our diagrams, a slight decrease in the tail length of the tail to the southward. But the amount is entirely too small upon which to distinguish a variety, much less a species, and the same thing holds true in regard to the scutellation. I have also stated that while a decrease in the scutellation in the southern parts of New Mexico and Arizona is VARIATIONS OF GARTER-SNAKES. 55 suggested, the tendency, if it exists, is not at all well defined. As regards the number of subcaudals and tail length in Arizona, the num- ber of available specimens is far too small to settle this point, although the maximum number of subcaudal plates in the males is 86, while the tail length remains the same as in the Chihuahuan specimens. As regards the number of ventral scutes there is an apparent reduc- tion, since the maximum number, even in the males, is but 165, while the minimum exhibited by the females is 153. It is to be observed, however, that although chiefly male records of the number of urosteges are to be had, this is due to the fact that it is the tails of the females in this lot that are mostly injured, and that the females actually predominate in the records, which explains the low number of ven- trals. In respect to the labials the average at Tucson is 8.3, which is practically the same as in Chihuahua. The similarity in scutel- lation and tail length between Arizona and Chihuahua specimens is thus very close, and there is no reason to believe that a larger number of specimens will materially disturb it, so that it is impossible to divide the group here on this basis. One more trait of the Mexican megalops remains for us to examine. Nearly the same difficulties have arisen over the attempt to dis- tinguish different forms within this species on the basis of color as we have seen above to have arisen in the attempt to divide it on the basis of tail length. As is well established by the specimens, there is a considerable range of variation in color, but it remains to be seen whether this is individual, sexual, or geographic. In general the color may be defined as some shade of olive or brown, with three light stripes distinct or obscure. The lateral is on the third and fourth rows, anteriorly, posteriorly on the third only,on the second and third, or indistinct. The dorsal stripe covers a varying distance of from 1 to 3 scale rows on either side of the median dorsal one. The lateral spots are obscured on the scales in the dark specimens, but are distinct in the lighter ones, and when present occur on all but the keels of the involved scales. Of the variations of this general type four color phases may be noted: First. The ground color is dull brownish olive, with the lateral spots visible on the edges of the involved scales, but not prominent, owing to the dark color. The lateral stripe is greenish olive, the dorsal stripe dull greenish yellow on the median and one-half of the adjacent rows, the edges not being well defined. The supralabials are dark greenish yellow, narrowly margined with black; the head olive brown and the parietal spots very small. The belly is bluish green, clouded with yellow anteriorly; the chin and throat also being vellow. (1279, Field Museum.) Second. The ground color of the above type may become so dark as to appear nearly black, entirely obscuring the spots on the scales. 56 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. The lateral stripe is then dark and nearly obscured, the dorsal stripe. represented only by dark yellow keels on the median rows of scales, the head black, parietal spots lacking, labials very dark brown, belly dark bluish, chin and throat yellow, and the supralabials brownish. (1097, Field Museum.) Third. The ground color is dark yellow, the black spots on the skin also conspicuous on the scales, and covering all but the keels of those involved. The stripes are only indicated by the fact that the spots do not encroach as much upon the scales of certain rows. The dorsal stripe is wide, covering from three to five rows. The head is brownish yellow, parietal spots very small, labials dark yellow and narrowly margined with black, the belly, chin, and throat bright creamy yellow. (1098, Field Museum.) Fourth. The ground color is light brownish olive; the lateral spots occur only on the edges of the scales, and the stripes are bright yellow and very distinct. The head is light brown, the parietal spots small, and the labials bright yellow, as are also the pre- and postoculars, and a postoral crescent. Distinct nuchal blotches. The belly is greenish yellow, and the chin and throat bright yellow. (1320, Field Museum.) tre Kennicott’s type of macrostemma corresponds in color to that which we have described as phase 1. In 1866 Professor Cope (1866, 306) described a new species which he called flavilabris, from two specimens sent to the Smithsonian Institution by Doctor Sartorius, which were labeled ‘‘Tableland or Southern Mountains of Mexico.” Unfortunately the types are now lost, but according to the descrip- tion there can be little doubt but that they were identical with color phase 4 as described above. There are at the present time three specimens in the U. S. National Museum (24993-4-5) which are labeled ‘‘South Mountains or Micrador Vera Cruz,’ and were col- lected by Doctor Sartorius. These are probably the specimens referred to by Cope (1866, 307), which, although he does not mention the resemblance, also seem to be referable to this phase. In 1885 Cope (1885b, 172) again established a new species (insig- niarum) on the basis of a specimen taken by himself at Chapultepec, Mexico, giving as its principal characteristics the absence of the dorsal stripe and occipital spots. It is impossible to distinguish the particular form he had in mind from this description, but fortunately both the type and cotype are available and correspond to our color phase No. 3; the cotype especially being a very good example of this style of coloration. In that paper (p. 173), in an attempt to give compara- tive diagnoses of the forms of this group, he distinguishes insigniarum from flavilabris by the absence of the dorsal stripe and occipital spots in the former. It will be seen, however, that this can not be done, since there is a dark phase (No. 2) which is similarily characterized VARIATIONS OF GARTER-SNAKES. 57 by an obscurity of these markings. In 1892 he apparently recog- nized this fact and the near relationships of these forms, as he describes them all as forms of macrostemma, evidently including insigniarum specimens with obscure markings both of the yellow (3) and black (2) phases, since he states that the type of macrostemma is intermediate in color between insigniarum and flavilabris. In 1900 he carried this farther and reduced flavilabris and insigniarum to the standing of varieties, but defined macrostemma (insigniarum) as being larger, darker colored and having the spots and bands indistinct and the parietal spots generally absent, as distinguished from flavilabris with its brighter colored ground, yellow labials, stripes, belly, and parietal spots. As we have shown in the description of the color phases, the ground color in the original insigniarum form is yellow, so that although Cope was justified in combining insigniarum and macrostemma, as all intermediate color phases occur, he has limited his description to include only the dark forms, phase 1 and 2, and placed the name insigniarum in the synonymy of macrostemma, but at the same time excluded the color phase to which it was originally given, putting these specimens with flawilabris. That this is actually the case is further shown by the fact that he subsequently labeled several specimens of the yellow phase (3) as flavilabris. As a matter of fact itis absurd to attempt to distinguish subspecies on these color phases, and, indeed, impossible to do so and still observe geographic probability. If we ignore all questions of nomenclature and examine the color phases, it will be found that numbers 1, 2, and 3 may all occur in the same locality and intergrade perfectly, but seem to be quite distinct from phase 4, with its more distinct markings. If we examine the range of the specimens, however, we will find that the color phases 1, 2, and 3 are only represented in the southern part of Mexico, and that they include all of the specimens taken about the lakes in this region (Patzcuaro, Chaleo, Xochimilco), - while phase 4 is found to the north in Arizona, Chihuahua, Durango, and Guanajuato, and in Veracruz and Puebla. They can therefore hardly be classed as either individual or racial variations, and since they are not sexual, males and temales being found of either color, the suggestion arises at once that they may be due more or less to local environmental influences. This explanation is enforced by the fact that the specimens taken about the lakes Chaleo and Patzcuaro (the forms with obscure markings) represent the aquatic forms whose habits have been described, while those in the more arid regions are the phase 4. This explanation must be taken with extreme caution, however, for in very few cases is there data with individual specimens. It is advanced principally to call attention to the need of detailed study on this point. Enough has been said to establish the homogeneity of the form megalops throughout its range. This uniformity, which is in har- 58 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. mony with the absence of geographic barriers, is expressed in scutel- . lation, tail length, and color. There are, however, variations both in scutellation and color. The color varieties are not individual or sexual, but seem to be partly at least associated with tne dampness of the environments. The variations in scutellation and tail length are in the nature of a scarcely perceptible decrease toward the south. This reduction, as shown by our diagrams, is a general one, and involves the number of dorsal scale rows, ventral and subcaudal plates, and supralabials, being best marked in the case of the dorsal scale rows. A ffinities.—If we can base our faith upon the position of the lateral stripe there can be no doubt that megalops is a member of the Radix eroup, since there is never any doubt of its position being upon the third and fourth rows, for the greater part of the length. Although posteriorly this stripe seems to descend to the second and third rows, there has not been a single specimen examined or recorded in whica it departed from the third and fourth rows anteriorly, even in those specimens with a reduced scutellation. It is in this respect sharply defined from all other forms on the plateau of Mexico, only to the northward (soutbern Arizona and New Mexico) coming in contact with a similar form. MARCIANUS.4 Description.—We have seen that the arid basins of southern Ari- zona and New Mexico are inhabited by one species of the Radiz group, the Mexican megalops. It is also the home of another char- acteristic and well marked form, known as marcianus, which is prob- ably entitled to the distinction of being the most pallid form in the genus. This species was described by Baird and Girard in 1853 (1853, 36-37) from a specimen taken near Cache Creek in what is now Okla- homa, and is characterized by the presence of 21-19-17 dorsal rows of scales, 8 supralabials, about 155 ventral plates, about 68 subcau- dals, and a tail length of about .23 in the males and .22 in the females. The lateral stripe is very light, on the third row of scales only anteriorly, posteriorly being upon the second and third rows. The general ground color is light brownish yellow, which is in marked contrast to the black markings. In view of the fact that it has for years been considered a subspecies of elegans, it may appear strange to some herpetologists that this species should be considered a mem- ber of the Radix group. We hope, however, to be able to justify our position by showing that there is in reality less difficulty in refer- ring it here than to the other groups in the genus. « Thamnophs marcianus (Batrp and Grrarp), Catalogue of North American Rep- tiles, 1853, pp. 36-37. Includes the Hutwenia elegans marciana of later writers, and EL. nigrolateris Brown (1889, pp. 421-422). VARIATIONS OF GARTER-SNAKES. 59 Habits and habitat relations.—Rarely is it possible to obtain a suf- ficient number of records to map in detail the range of any form, but if we know its preferred habitat we can, in doubtful regions, determine more or less accurately the probable extent of its range by the limits of the environmental conditions with which it is usually associated. In the present case such data would be of great value, but unfortunately, as far as we have been able to find, there is prac- tically nothing recorded on the habits of marcianus, with the excep- tion of the general conditions of the region it inhabits. I have observed elsewhere (Ruthven, 1907, 589) that it is occasionally at least found in the vicinity of streams, which is in harmony with the known habits of the other arid region forms in the genus (fig. 20). Fig. 20.—SANTA Cruz RIVER AT TUCSON, ARIZONA. THAMNOPHIS MARCIANUS HAS BEEN FOUND HERE. Range.—The region inhabited by marcianus includes the Proplateau region of southern Arizona, New Mexico, and Texas, the central part of the latter State, and northern Mexico. We have already briefly discussed the geographic and climatic conditions of the Proplateau region, so there remains to be considered only the environmental conditions in the Texas portion of the range of this form. This region may for convenience be divided into two physiographic regions; the prairie-plains region, situated roughly to the east of the Pecos River, north of the thirtieth parallel and west of the ninety-eighth meridian, and the Rio Grande plain, which occupies the triangular area between the prairie-plains, Rio Grande River, and ninety-eighth 60 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. meridian, As here defined, the prairie-plains region includes the. ‘‘Central Province” of Hill and the strip of prairie that extends southward from Austin to the Rio Grande plain. The Rio Grande plain, as we use the term, is synonymous with the ‘“‘Lower Rio Grande Country” of Hill (see fig. 21). ~To the east of the East Front Ranges, the western Texas region descends from the high plateaus bordering the Rocky Mountains in northern New Mexico by a series of broad plains arranged in the form of a great stair (as described by Hill) to the Gulf of Mexico. The higher part of this area constitutes part of the plateau region of western United States and is bounded on the south and east by a 4 ALS A ILS: Fic. 21—THE NATURAL DIVISIONS (ENVIRONMENTAL COMPLEXES) OF THE GREATER TEXAS REGION. line running from Del Rio to Austin and from here west and north to Oklahoma. To the eastward the plateaus grade down into the prairie region, which forms a broad band across central Texas and extends to the northward beyond the State. The climatic conditions of these regions are well illustrated by the character of the vegetation. The Trans-Pecos region is a part of the Proplateau region of southern Arizona and New Mexico, and the conditions are very similar to those in the latter States. The rainfall is mostly below 10 inches, and the evaporation is high, which with the scanty rainfall prevents the occurrence of arboreal vegeta- VARIATIONS OF GARTER-SNAKES. 61 tion, except on the higher summits. The flora of the plains and the slopes below the timber zones consists of such forms as the sotol, ocotillo, creosote-bush, mesquite, and a host of cactuses. To the east of the East Front Ranges the climate becomes pro- gressively more moist than that of the Proplateau region, although the rainfall continues below 30 inches to the ninety-eighth meridian. On the sandy plains of the lower Rio Grande, with its dense growths of chaparral and intervening prairies, the conditions are still very arid, owing to the low humidity and small rainfall, but to the eastward the grasses become more mesophytic and grade into the eastern for- ests near the 98th meridian. To the northward the forest margin is bordered on the west by the mesophytic grass associations of the prairies that rise to the westward to the solid buffalo grass forma- tion of the high plains. The ninety-eighth meridian, as has been pointed out by Hill (1900) and in more detail by Bray (1901 and 1904), and Bailey (1905), marks in a general way the boundary between the mesophytic forest of southeastern United States and the great grass country of Texas. This doubtless marks equally well the dividing line between the fauna of these two areas. As Bailey has shown, it marks the eastern limit of the mesquite, Texan woodpecker, and Texas rattlesnake. Owing to the extreme paucity of records, it is impossible to draw the geographic boundaries of marcianus, except in a very general way. The range has been defined by Cope (1892, 656) as ‘ ‘restricted to the valley of the Rio Grande from Colorado to its mouth, extend- ing eastward into Texas as far as the Concho and Nueces rivers,”’ but as specimens have been taken in Arizona, the range as given by Brown (1901, 24), ‘‘Central Texas to western Arizona’’ is more nearly correct. But little importance can be placed in locality records for this form until the specimens have been examined. In 1875, Yarrow (1875, 573), in writing of this form, remarked that the last specimens of vagrans (elegans) seen were at Taos, New Mexico, on the north side of the Picoris Mountains, while on the south side of these mountains marcianus was said to occur for the first time, and specimens were listed from ‘‘San Ildefonso, New Mexico,” ‘‘Abiquiu, New Mexico” (3), ‘‘Taos, New Mexico,” and ‘‘Pueblo, Colorado” (2). Unfortunately only field numbers are given for these specimens, so that they cannot be located with certainty, but the only speci- mens in the U. 8. National Museum from these localities bearing the date and collectors recorded by Yarrow are three specimens of ele- gans from Abiquiu (No, 8728), one specimen of elegans labeled ‘‘New Mexico” (No. 8421), and two specimens of radix from Pueblo (No. 8581). The particular specimens from San Ildefonso listed by Yar- row as marcianus cannot be determined with certainty, as there are a number of specimens in the U. 8. National Museum bearing the same 62 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. date and locality. There are, however, two specimens from San Ilde- . fonso (8416-8417) of eques that are labeled marcianus, while it is to be noted that the only specimens from Taos in the Museum are elegans. With these gross errors confronting us in the literature, it is evi- _ dent that we can trust neither the general ranges given nor the detailed locality records. I may be excused, therefore, for confining my discussion of the range of this group principally to the specimens that I have examined. Fortunately the doubtful localities are rare, for I have examined the specimens upon which most of the records have been based. J have examined specimens with defi- Fig. 22.DISTRIBUTION OF THAMNOPHIS MARCIANUS, AS INDICATED BY THE LOCALITY RECORDS. nite locality records as follows: Fort Yuma, California; White Horse Springs, and Fort Supply, Oklahoma; Matamoras and Charco Escon- dido, Tamaulipas, and Ojo del Diable, Chihuahua, Mexico; Browns- ville, San Diego, Point Isabelle, Reutersville, Cameron County, San Antonio, Indianola, Eagle Pass, Pecos, San Angelo, Helotes, Jeff Davis County, ‘‘South of Clarendon,” Kerrville, Texas; Tucson, Arizona; ‘“Red River, Arkansas. ”’ When these localities are plotted on a map (fig. 22) the distribution of the form is seen to be more extensive than has been generally stated. In the Proplateau region it is known from as far west as Yuma, VARIATIONS OF GARTER-SN AKES. 63 Arizona, but, as has been elsewhere stated (Ruthven, 1907, 589), since only three ® specimens have been recorded from here and this is the only record for the Sonoran desert region in Arizona, they should not be taken as establishing the occurrence of the form so far to the westward, for the locality may not be an exact one. Its northern and southern limits in this State can only be conjectured. The single specimen in the U. 8S. National Museum labeled ‘‘Ojo del Diable, Chihuahua,” is the southernmost record for the Proplateau region. Aside from the Yuma records, which consist of one specimen in the Academy of Natural Sciences of Philadelphia and two in the U. S. National Museum, all collected by Major Thomas, no other specimens are known to me from southern Arizona, except three in the Academy of Natural Sciences of Philadelphia, two in the National Museum, and one in the American Museum, all from Tucson. It is probable, however, that it ranges north to the Colorado plateau, although no specimens have been taken at Fort Whipple, where con- siderable collecting has been done. It is highly probable that it does not range north of this point. For New Mexico there are no authentic records. As we have seen above, Yarrow’s specimens are to be referred to elegans, eques, and radix. Cope (1883, 12) records a specimen taken by Frank Snow at Socorro, but I have been unable to verify this record. Without doubt marcianus occurs in the southern part of this State, but its northern limit is a debatable point, for although the high plateau, as in Arizona, may exclude it from all but the extreme southern part, on either side of the Rio Grande, there seems no good reason why it should not occur up the valley of this river at least as far north as Las Cruces. In the Trans-Pecos region of Texas marcianus has been recorded -from the Davis Mountains, Jeff Davis County, Paisano, and Boquil- las, all localities south of the high plateaus. Kast of the Kast Front Ranges the records indicate that it occurs throughout the prairie region of central Texas, and the Rio Grande plain as far south as Charco Escondido, Tamaulipas, north to Fort Supply, Oklahoma, and westward to an undetermined distance on the plains. Claren- don, Pecos, and San Angelo are the most western records for this region, and it will be noticed that these localities mark approxi- mately the eastern margin of the Staked Plains. Whether this indicates the actual western limit of the form in this region, or the lack of specimens from more western localities, can not of course be determined. As in the Proplateau region, however, it is signifi- cant that its chief distribution in Texas is confined to the lower altitudes, so that it would not be surprising to find that its western range is limited by the increasing altitude of the high plateau, the Pecos records possibly indicating that it pushes to some extent up a Erroneously given as two in the paper mentioned. 64 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. the river valleys. The eastern limit of its known range is repre-. sented by the Victoria, San Antonio, and Austin records, and is approximately the 98th meridian, which, as we have seen, marks the boundary between the prairie type of biota of central Texas and the forests of southeastern United States. The most northern record is Fort Supply, Oklahoma (see p. 69). The range of this form then may be defined as the arid deserts of northern Mexico, southeastern Arizona, southern New Mexico, and southwestern Texas, the arid plains about the lower part of the Rio Grande, and the prairie region of central Texas. Variation.—The individuality of marcianus is so pronounced that it stands out in strong relief from the other forms in the region which it occupies, and makes the question of its affinities a very puzzling one. At the same time it is a very homogenous form and apparently varies but little either geographically or individually, as shown by the fact that but one other variety (nigrolateris, Brown, 1889), based on an anomalous specimen, has been made from it, a form since dropped. Any discussion of the variations of marcianus must be made with ereat caution in consequence of the inadequate number of speci- mens available. Although the material in the American, Field, and U. S. National Museums, and the Philadelphia Academy of Natural Sciences have been examined, besides a number of specimens from private collectors, it has been impossible to obtain more than 60 individuals of this form. Fortunately these appear to be scattered over most of the range of the species, and this with the apparent lack of marked individual variation makes it possible, even with but a few specimens from the different localities, to examine the geographic variation. In every specimen examined from Tucson, Arizona, to Oklahoma, the scale rows are 21-19-17. The supralabials are more variable; at Tucson they are constantly 8 in 5 specimens; at San Antonio, 8 and 9; in Cameron County they are 8 in the great majority of cases, occasionally 7, and in one case 9 on one side, with an average of 7.9; in Oklahoma 8. — Evi- dently from the material at hand it is impossible to say that there is any geographic variation in this character. As regards the ven- tral plates, the females have 151 to 156, in the specimens labeled “Yuma, Arizona,” and the males 157 to 162, with an average (all specimens) of 155.6; at Tucson two males have 160 and_ three females 156-159, average 158.4. At San Antonio a female has 149 and a male 157; while in Cameron County the range in variations of both males and females are lowered, in only two cases reaching 160 and falling as low as 144, the average being 151. At White Horse Springs, Oklahoma, a male has 157 and a female 155. ‘These figures seem to indicate a slightly larger number of gastrosteges in VARIATIONS OF GARTER-SNAKES. 65 Arizona and a decrease in the Rio Grande Plain, but, while the average for Cameron County probably approximately indicates the actual conditions, the other averages are based on far too small a number of specimens to be reliable. The most striking fact shown by these figures is the uniformity in the number of ventral plates throughout the range of the group, for although enough records are not at hand to reveal any slight trends of variation if they are present, enough have been secured to indicate beyond question that such variations, if they actually exist, must be small, as the averages from the different localities differ less than the amount of individual variation in any set of records. The evidence of the sub- caudal plates is especially unsatisfactory, for it is evident that not nearly enough records have been obtained to determine even approximately the range of sexual variation. The specimens labeled ‘‘Yuma” vary from 62 to 71, average 66.6. At Tucson the average is 68.6; in Cameron County the average is 68.6, while at White Horse Springs a male has 71. Here again we find that the averages when males and females are combined falls between 66 and 69, a difference of but 4 scutes and one that could easily be due to the predominance of either sex in the averages. As in the case of the subcaudals, it is impossible from the mate- rial to determine the limits of sexual variation in the tail length, yet an examination of the measurements indicates that the males, as a rule, vary from .22 to .26, the general length being between .23 and .24. The females seldom vary more than two points about .21, the range being from .19 to .22. It is not to be inferred from the above discussion that there is no geographic variation in the form, for it is possible that when a large enough number of specimens have been examined to establish the limits of sexual variation it will be found that there is some varia- tion in different parts of the range. But nevertheless the evidence now available indicates that if such a variation is present it is small, for even this small amount of material, subject as it is to irregular- ities caused by individual and sexual variation, demonstrates unques- tionably a close uniformity in scutellation and tail length throughout the entire range. In color this uniformity is even more strongly enforced, and can be no more plainly expressed than by the statement that, owing to the constancy of coloration throughout the extent of its range, no varia- tions have been considered of specific or subspecific value. The ground color is usually light brownish yellow above, with three rows of alternating black spots; the first row on the first, second, and occa- sionally part of the third rows of scales, the other two rows alternating between the lateral and dorsal stripe, on the skin and all but the keels 66 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. of the involved scales. The stripes are very light yellow; the lateral stripe on the third row only anteriorly, and posteriorly on the second and third; the dorsal on the median and about one-half of the adjacent rows encroached on to the median row by the upper row of spots. Large nuchal blotches and a distinet postoral crescent. Head dark olive; first, second, third, seventh, and eighth supralabials olivaceous, the fourth, fifth, and sixth, with the preoculars and second and third postoculars, being light yellow. The third, fourth, fifth, and sixth supralabials margined with black. Belly light yellowish ash. In older specimens the ground color becomes darker, and the spots seem to retreat from the scales, although still distinct on the skin. The spots below the lateral stripe also become indefinite, but the stripes themselves retain their position. This variation, however, is appar- ently due to age, as we have said, and as far as we have been able to see there is no marked geographic differences in color. A ffinities—Considered in the totality of its characters, marcianus stands out in decided contrast to the other forms in this region, and, as we have stated before, this individuality has in great part delivered it from the confusion that has resulted in the efforts to classify many of the other forms. The attempts that have been made to determine the relationships of marcianus have, however, led to several extraor- dinary results. It was described in 1853 by Baird and Girard (1853, 36-37) as a distinct species and no attempt was made to establish its affinities for thirty years. In 1883 Garman (1883, 25 and 138) included it as a subspecies of sirtalis, which, however, means little as regards genetic relationships, for we find this writer including as varieties under this species such divergent forms as radix, elegans, and sirtalis. Cope in 1892 (1892, 656) reduced it to the rank of a variety in the Elegans group, on the basis of the position of the lateral stripe on the second and third rows, and the possession of 21 seale rows and 8 upper labials, thus allying it to elegans and hammondi. This arrangement was followed by Brown in 1901 in his review of the genus (1901), but in 1904 (1904, 470) he changed his mind and derived it directly from parietalis, with no explanation as to his reasons for so doing. These dispositions of the form are unsatisfactory, to say the least. It will be observed that the placing of marcianus in the Elegans group rested upon the assumptions that the lateral stripe is to be considered on the second and third rows of scales, which is far from established, and that racial affinities are indicated by similarity in the number of scale rows. The latter, as we have seen, is also far from being the case, and we are thus freed from the necessity of accepting any of the solutions of the problem that have been based on these grounds. We must search for similarities with other forms : VARIATIONS OF GARTER-SNAKES. 67 in the light of our knowledge of the method of variation and of geographic probability. The only character, so far as I can see, that would justify an attempt to derive this form from parietalis and relate it to eques is the pallid color. In general the ground color is nearly the same as in eques, while the paleness of the labials and stripes, and the similarity in the arrangement of black markings, is so close as to result in frequent confusion. Color, as we have seen, is the least important of our criteria, but it is worthy of note that in spite of the pallidness in eques the lateral stripe is always distinctly on the second and third rows throughout the entire length (instead of on the third row only), even in the region in which it occurs with marcianus. In scutellation there is very little resemblance between marcianus and parietalis or eques. The latter in Texas has mostly 8 supralabials and always 19-17 scale rows, and parietalis has usually 7 supralabials and always 19-17 scale rows. According to our observations on the variation in the number of scale rows, marcianus could have been derived from either parietalis or eques by an increase in the number of scales, but the further difficulty presents itself that both of the latter species (especially eques) overlap the range of marcianus considerably with- out effecting the purity of the types, and the only way by which they can be considered akin is to invoke the aid of some form of isolation (ecological or physiological), which the evidence of similarity does not call for. The difficulty of adopting such an explanation would be less if there were any signs of intergradations in these forms; but, as it is, marcianus has constantly 21-19-17 rows and eques and parietalis 19-17. Its relation to the Elegans group is not, however, so easily disposed of. In the possession of 21-19-17 scale rows and 8 labials marcianus agrees closely with hammondi and elegans, and while the number of ventral and subcaudal plates is normally less and the tail shorter, the inference might well be drawn that the similarity is an expression of relationship. Nor does the lateral stripe confute this, for it is the general pallidness of the ground color that apparently makes the stripe indistinct except upon the third row. Geographically also it seems, at the first glance, as if this relationship was confirmed, for the range of hammondi, elegans, and marcianus seem to be exactly contiguous, and do not overlap. There are certain facts, however, that confute the evidence furnished by the similarity in scutella- tion. First and foremost, is the lateral stripe to be considered as being upon the second and third rows? Secondly, hammondi is markedly distinct in color from marcianus even in the regions where they approach each other, and this distinctness is supported by the general topography of the body, as expressed in a longer, narrower head, increased number of urosteges, etc., and the presence of two 68 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. preoculars. Very evidently it seems to us that if we grant arelation- ship between these forms it must be a distant one, for there can be no question as to their distinctness at present. As regards elegans, the geographic probabilities may also be con- ceived as favoring a close relationship, for it seems very evident that in Arizona, New Mexico, and Texas, where the range of the two forms come together, elegans is restricted to the plateaus and mar- cianus from them, and the differences between the two forms may be explained by this separation of their respective ranges and as the result of the different conditions to which they are exposed. Thus elegans, even in the most southern localities from which it is known, San Ildefonso, New Mexico, presents so often a scale formula of 19-21-19-17 as to indicate plainly a tendency toward a reduction in scutellation, which may possibly be correlated with the higher altitude of its habitat, while in coloration it resembles so closely dark specimens of marcianus as to make it difficult at times to refer specimens to the proper form. Furthermore, the generally light color of marcianus, as I have shown, may be correlated with the aridity of its habitat. What, then, is, the objection to this attractive view? If we take for granted that elegans is a dwarfed form of a stock that possessed a larger number of scale rows, it seems to us that we have reason to expect that correlated with the decrease in the number of scale rows there at least will not be an increase in the number of ventral plates, although the tail length and urosteges may vary independently, as we have seen. If the reduction in the different series of scales on the body is correlated, as seems to be the case, then elegans has apparently been derived from some stock with a larger scutellation than marcianus, since it has a decidedly larger number of ventral plates. This problem will be taken up again when the Elegans group is considered, but it seems to me, although the material is as yet much too meager to warrant any decided opinion, that the neighboring species of the Hlegans group all give evidence of being derived from a different source than marcianus, and the opposite. If this point be granted, I have now excluded marcianus from all except the Radix group, and it remains for me to examine its relations to the other forms of this group. The fact that the first row of spots covers the first and second rows anteriorly seems to furnish some evi- dence that the stripe is to be considered on the third and fourth rows. Still, this is, of course, not conclusive, and the lateral stripe must still be considered noncommital, although the fact that it occurs on the second and third rows posteriorly is not an insurmountable objection, as it also has this position in megalops, in which it is anteriorly upon the third and fourth rows. The number of labials is much the same in both marcianus and megalops, as is also the tail length. On the VARIATIONS OF GARTER-SNAKES. 69 other hand, the number of dorsal scale rows, subcaudals, and ven- trals are apparently less. Thus there are no serious objections to deriving marcianus from megalops on structural grounds (the geo- graphic probilities will be considered later), but still there is no satis- factory proof that such a relationship actually exists, and we must look elsewhere for evidence of its relation to the Radix group. On the north marcianus meets radix, and when the scutellation of these forms is examined it is seen that there is no great break between the forms in this regard. The number of scale rows and labials are the same near the common boundary, and, notwithstanding the fact that marcianus has a smaller range of variation in the number of ventrals (157-163) as compared to radix, 159-172, in the same general region, the averages are very close, while the tail length and number of subcaudals are approximately the same. In spite of the fact that the color of radix is usually darker, the general pattern is very similar to that of marcianus. Indeed, in western specimens of radix the ground color becomes quite light, so that in the case of two specimens from Oklahoma, in the possession of the Field Museum (Nos. 630, 631), it is impossible to tell whether they are to be referred to radix or mar- cianus. In these specimens the color is dark brownish yellow, the first and second rows light ash. The spots are in three very distinct, alternating rows on the skin and all of the involved scales; the first row being on the first and second rows, the other two between the stripes. The lateral stripe is distinct, narrow, and anteriorly on the third and part of the fourth rows, posteriorly on the second and third. Anteriorly the spots tend to fuse into large blotches. Eye rather small. Head dark olive. The first, second, seventh, and eighth supralabials are more olive, the third to sixth, inclusive, being light yellow; all are margined with black, the fourth, fifth, and sixth the heaviest. This coloration is so like that which exists generally in marcianus as to make it evident that these specimens are closer to _this form than any other. The points that indicate an affinity with radix are the fact that the first and second rows of scales differ in color from the lateral stripe and are covered by the first row of spots, and that the lateral stripe seems to be partly on the fourth row. If I am right in considering that these specimens from Oklahoma indicate. that marcianus and radix are directly related, the lateral stripe in the former must be considered as belonging on the third and fourth rows of scales, which throws the form with the Radix group. Whether marcianus and radix intergrade at the present time along their common boundary, or not, is not a question to be decided on the basis of two specimens, but, be this as it may, it must, we think, be conceded that there is such a very close similarity in scutellation and color between the two forms as to warrant the conclusion that they are directly related, while if we consider the lateral stripe in the former 33553—Bull. 61—08——6 70 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. as a modification of the third and fourth row position, there is little difficulty in placing marcianus in this group. FAS AGS: Fic. 30. DISTRIBUTION OF THAMNOPHIS BUTLERI, AS INDICATED BY THE LOCALITY RECORDS. limit of its occurrence. These specimens will be further discussed in the consideration of the variation of the group. At this point mention should again be made of the three specimens in the U.S. National Museum from Milwaukee, Wisconsin. I have given my reasons for referrmg these specimens to radix, but it must be admitted that they resemble butleri so closely as to be indistinguish- able, and it may be that further material will show them to belong to this form and extend the range around the southern end of Lake Michigan, but the solution of the problem given on page 85 seems the most plausible one at present. It is impossible with our imperfect knowledge of the limits of its range to point out any close relations between the distribution of a As this paper is passing through the press I have received a specimen of butleri from Dayton, Ohio. 94 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. butleri and the conditions with which it is found associated. It must be acceded, in a general way, however, that, as far as our present knowledge goes, its range is apparently included in the area lying east of the prairie peninsula, which is characterized by the occurrence of open torests on the uplands (fig. 30). If, as I hope to be able to . do, I can show that butleri is genetically related to radix this dis- tribution may be easily accounted for by the fact that as butlert has prairie affinities, it might be expected to prefer conditions more in accord with those of the treeless region. I have already shown that the conditions in this region are somewhat intermediate between those of the prairie and the denser forests of southeastern United States, and, if a direct relationship exists between butleri and radix, the coincidence between the range of butlert and the open forest area may be accounted for by the affinities of both. This subject will be reverted to in discussing its affinities. It should be noted, however, 19-17 17-49-17] Bids ae eslbs, Fic. 31.—DIAGRAM SHOWING THE VARIATION IN THE DORSAL SCALE FORMULA IN THAMNOPHIS BUTLERI. Indiana. Michigan. Ohio. Pennsylvania. that the range of butleri is entirely within the glaciated area of eastern North America which would indicate that the form is of post-Glacial origin. Variation.—The characters of butleri are quite constant and have not by their variability given rise to much confusion. The scale rows are invariably 19-17 in nearly all of the specimens examined, but it should be observed (fig. 31) that while a larger formula has never been observed, three of the Michigan specimens have the for- mula 17-19-17 and both of the Pennsylvania (brachystoma) specimens have 17-19-17-15, the smallest number observed. According to Cope, the type of butleri has 7 superior and 8 inferior labial plates, a formula that holds for the Lake Makinkuckee, Miami River, and Sandusky specimens (figs. 32 and 33). The Turkey Lake specimen has the same number, except on the right side, where one superior is dropped to leave 6. At Ann Arbor and Olivet, however, 6 supra- VARIATIONS OF GARTER-SNAKES. 95 labials occur in practically one-half of the specimens, and, while the infralabials are occasionally 8, 9, or 10, the average is very close to 8. The single specimen from Columbus has 6/8, the one from Frank- Indiana. Michigan. Ohio. Pennsylvania. Fic. 32.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUPRALABIALS IN THAMNOPHIS BUTLERI. : Indiana. Michigan. Ohio. Pennsylvania. F1G.33.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF INFRALABIALS IN THAMNOPHIS BUTLERI. lin, Venango County, Pennsylvania has 6/8, and Mr. Fowler’s speci- men from McKean County, Pennsylvania, 6/6-7. An examination of the ventrals shows the following distribution: Ann Arbor, males 134-146, females 132-140. All of the Indiana 96 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. and Ohio specimens fall within these extremes, as do also both of the Pennsylvania. Subcaudals (Ann Arbor, Michigan), males 61-66, females 49-58. The male records are based on but few specimens, however, and the number probably ranges both lower and higher. Indeed, a male from Columbus shows 59. The type of brachystoma, although doubtful, is probably a male and has 72. In the McKean County specimen the tail is injured. The series are too small to reveal geographic differences, if they exist. One must be cautious in drawing conclusions from such a small amount of data, but the fact that in very few specimens west of Pennsylvania does the reduction in the dorsal scale formula become so low as in the Pennsylvania specimens nor the labial formula as small as in the McKean County specimen seems to us to indicate a reduction in scutellation to the eastward, at least as far as these characters are concerned. At any rate, it is evident, as is shown by the diagrams, that butleri as a form is characterized by a very reduced scutellation. Affinities.—The only form with which butler: has been confused is sirtalis. Its distinctness from this form has been pointed out (Ruth- ven, 1904, 295 and 298), and this distinctness is such that there is slight reason for deriving it from this form, from which it differs in the position of the lateral stripe, the number of labial plates, and color, while lying entirely within its range. It also differs from sauritus in the number of labial plates and in the tail length, and also lies entirely within its range. Here again the difficulty in establishing relationships lies in the fact that the position of the lateral stripe is noncommittal. It is true that posteriorly it is upon the second and third rows, but this is frequently the case in forms with the stripe on 3 and 4, when the fourth row is lost to leave 17 (megalops and marcianus). Anteriorly the lateral stripe is on 3 and one-half of 2 and 4, and it seems that this is probably due to a tendency to the loss of the fourth row, for where it is present the stripe is always partly upon it, while when it is absent the stripe is on 2 and 38. The only other form whose range it approaches is radix, and if the position of the lateral stripe be considered as a modification of the position which it occu- pies in this form (on the third and fourth rows), everything seems to point toward radix as the nearest relative; for, although radix has generally a larger scutellation, we have already seen that in the extreme eastern part of its range, which corresponds very closely with the western limit of butleri, a reduction occurs that brings the number of scale rows and labials, at least, exactly to the formulas that characterize butleri. I believe that I am justified in con- cluding, therefore, that this form is a member of the Radix group, and that its closest relative is radix, which it meets on the west. y, > eel ————— eee ee ee ee — VARIATIONS OF GARTER-SNAKES. 97 CONCLUSION. If the affinities of the different forms as indicated above (fig. 34) be accepted, this group, extending as it does from the Mexican plateau to the Great Lakes region, exhibits a progressive decrease in size and scutellation from the Mexican plateau toward the extreme limits of its range. The reduction is slight in each particular form, but when the extremes are compared with the center it becomes very apparent. Thus, butleri and megalops in northern Mexico are the extremes in the number of scales in each of the series which I believe to be correlated with size, while at the same time aslight knowledge of these forms is sufficient to show that there is also a striking difference in size between the two. It should also be noted that the areas in which the transitions from one set of scale formulas to another takes place radix — -butleri re marcianus mecalops 5 | p Fig. 34.—PHYLOGENETIC DEVELOPMENT OF THE RADIX GROUP. lie mostly between the forms, and are apparently very narrow, except between radix and butleri since radix shows a marked tend- ency toward a reduced formula throughout the prairie region. If the relationships are as I have described them, it should be noted (1) that in each region of different environmental conditions which it occupies the group is now represented by a different form; (2) that each form is characterized by a smaller scutellation than its neighbor toward northern Mexico, and exhibits its minimum scutellation at some point on the outskirts of its range, where it comes in contact with another form with the scutellation similar to that of the reduced individuals; (3) that the most dwarfed repre- sentative of the group is now found in the form farthest removed, geographically and genetically, from the northern part of Mexico, 98 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. THE SAURITUS GROUP. (PROXIMUS, SAURITUS, SACKENI.) PROXIMUS.4 Description.—Contrary to most forms in the genus, there is little danger of confusing proximus with any other garter-snake, The stripe is always upon the third and fourth rows and the form is long and slender, the tail forming .25 to .37 of the total length. The dorsal scale rows are nearly always 19-17 (in the southern part of the range occasionally 17-19-17) and the labial formula is usually 8/10, only occasionally 8/9 or 8/11, and more rarely 7/10 or 7/9. The ventral scutes vary between 150-179 and the urosteges between 75-125. In coloration proximus closely resembles the other forms of the group. The ground color varies from light olive to brown or black and is never broken up by distinct spots, although a narrow irregu- - lar black band is generally found along the stripes in specimens hght enough to show it and many of the scales may have black edges and bases. In some of the paler specimens examined the usual two rows of alternating spots are distinct upon the skin, but in by far the majority of specimens these spots are fused more or less, so that the skin between the scales is usually black with numerous white lines irregularly disposed. The stripes may be of a bluish, yellowish, or greenish tint (the dorsal rarely reddish, occasionally brown), and the lateral is in every specimen examined upon the third and fourth rows; the dorsal usually occupies the median and the halves of the adjacent rows, but occasionally covers nearly all of the adjacent rows. The superior labials are always quite pale, occasionally red- dish, in color and nearly always lack the black border that is found throughout all of the other groups. The ventral surface is gener- ally bright yellowish, greenish, or bluish, and without or with very small and mostly concealed ventral spots. Habits and habitat relations.—Very few records of observations upon the habits of this snake can be found. Under the name Eutez- nia saurita var. faireyi, Taylor (1892, 322) writes for Nebraska that the food ‘‘consists mostly of insects and their larve, but also includes fish, frogs, ete.’”’ Upon what authority this statement is made we do not know. In regard to three specimens taken at Progreso, Yucatan, Dr. L. J. Cole has written me as follows: ‘‘All of the speci- mens [three] were found in the water (brackish) in the mangrove swamps back of Progreso. One other specimen was seen, and that also was swimming in the water. I saw none of them on land.” Cope says (1880, 23) that it is ‘‘like the E. saurita, aquatic in its habits.”’ aThamnophis sauritus proximus (Say), Long’s Exped. Rocky Mts., 1823, p. 187. Includes Eutenia rutiloris Corr and LE. faireyi Barrp and GIRARD. VARIATIONS OF GARTER-SNAKES. 99 By far the best account of the habits is that of Ditmars (1907, 220-221): In habits this snake appeals to the two preceding species [sauritus and sackeni]. It is very quick in its motions, and appears to be perfectly at home in the water, swim- ming with agility and extreme grace and diving to the bottom of a pond or stream and there secreting itself among aquatic plants. Captive specimens are very hardy, and will live indefinitely upon a diet of small frogs or fishes. A number of specimens in the writer’s collection were very fond of climbing into a small branch that had been placed in their cage. Here they would coil in a tight cluster, with heads protruding in every direction. Upon the intro- duction of food they would dart for the prey in frenzied fashion, the lucky individ- uals thrashing their tails violently as if to distract the attention of their hungry asso- ciates from the morsels in the jaws of the former. One of these snakes gave birth to fifteen young on the 24th of August. While our knowledge of its habits is thus very meager, from what is known of the other forms in this group I believe that, like them, proximus will be found to be more aquatic in its habits than most of the other forms of the genus. Range.—Proximus is known to occur on the coastal plain from the eastern coast of British Honduras to the Mississippi River, and to the northward of this plain and west of the Mississippi River in the prairie-plains region and southeastern forest region of North America to about the latitude of the northern boundary of Iowa. The eastern coastal plain in Mexico is a low-lying tract of land bordering the Gulf of Mexico. The climate is hot and humid, the temperatures being tropical and the precipitation excessive. It is the ‘‘tierra caliente’ of the Mexicans, and is characterized by a rich tropical flora that forms dense jungles. In Texas and Louisiana the coastal plain conditions are similar to those in Mexico. Extensive tide- washed, brackish marshes occur all along the coast, and are accom- panied on higher ground by a biota of distinct tropical affinities (Bray, 1901, 102-103; Bailey, 1905, 16-18). The prairie region of North America has been briefly described. The southeastern forest region, which occupies southeastern United States south and east of the prairie region, is characterized by being the principal area of development of the hard-wood forest; it is the home of Quercus alba, Magnolia acuminata, Acer saccharum, Fagus americana, Liriodendron tulipifera, Fraxinus americana, Quercus rubra, and Hicoria alba. Transeau (1905) has shown that the region is characterized by a rainfall evaporation ratio of 100-110 per cent and that where this ratio falls to 80-100 per cent the flora merges into that of the prairie region. The southeastern forest region to the south of the prairie peninsula crosses the Mississippi and extends to the eastern limit of the prairie, which, as previously stated, is the western edge of the Ozark highlands and the 98th meridian. Proxi- mus enters this region, then, only in its western part. 100 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. Specimens of proximus with definite locality data have been exam- ined as follows: Progreso and Cozumel Island, Yucatan; Orizaba, Xalapa, and Tuxpam, Veracruz; Matamoras, Tamaulipas; Caderita, Neuvo Leon; Tule Canyon, Dallas, Matagordo, Pecos, Fort Cobb, Brownsville, Wheelock, Helotes, New Braunfels, Austin, Kerrville, San Pedro, San Angelo, Fort McKavett, Devils River, High Bridge, Pecos River, and Fort Stockton, Texas; New Orleans, St. James Parish, Belair, Slidell, Prairie Mer Rouge, Calcasieu Pass, Perry, and Grand Coteau, Louisiana; Greenway, Arkansas; Butler County, Montgomery County, St. Clair County, St. Louis, Missouri; Neosho Falls and Dora, Kansas; Nemaha County, Nebraska; Ames and Des Moines, Iowa; Chicago, Rock Island, Olney, and Mount Carmel, Illinois; Fox River and Racine, Wisconsin. Boulenger (1893, 214) records specimens from Belize, Honduras.* Bailey (1905, 48) gives the following Texas localities: ‘Brownsville, Lomita Ranch (Hidalgo County), Sycamore Creek, Corpus Christi, and San Antonio River, ncar San Antonio.’”’ Taylor states that he has examined specimens from Nemaha, Saline, and Saunders coun- ties, Nebraska. Branson writes that in Kansas he examined speci- mens from ‘“‘ Wallace, Douglas, Franklin, Geary, Woodson, Clark, and Shawnee counties.’’ At present we consider all of the Illinois and Wisconsin ribbon snakes as belonging to this form, but,-as later stated, more evidence may show that the specimens in this region are not typical. Ribbon snakes occur throughout the former State,? and in Wisconsin at least to Racine, and probably farther north, for a single specimen in the U. S. National Museum (No. 731) is labeled “Fox River.’’ No definite Minnesota record has been found, the only record being a single specimen (No. 6179) in the Academy of Natural Sciences of Philadelphia, labeled ‘“ Minnesota.”’ The distribution of this form (fig. 35) is thus in harmony with its aquatic habits. In Mexico it is confined to the coastal plain, being apparently unable to push out from this area. In eastern Texas, however, it encounters the eastern forest, which, as would be expected, apparently furnishes conditions somewhat similar to those of the coastal plain and permits the form to extend its range throughout this environment west of the Mississippi, and even to extend entirely across the prairie region and into the great plains and northeastern forest regions along the streams. Detailed evidence is not at hand that provimus is confined to the vicinity of streams in the prairie 4 Boulenger (1893, 213) also records a specimen from the Atoyac River, Guerrero, and Giinther (1894, 132) gives the locality Guatemala, but as these localities are the only ones outside of the eastern coastal plain of Middle America, the data and identifi- cation of the specimens should be ascertaincd with certainty before the range is extended to include them. ; > TH. Garman (1892, 265) gives the following Ilinois localities: Chicago, Cook County; Peoria, Normal, Jersey County; Mount Carmel, Union County. el er rh VARIATIONS OF GARTER-SNAKES. 101 Fic. 35. —DISTRIBUTION OF THAMNOPHIS SAURITUS PROXIMUS, AS INDICATED BY THE LOCALITY RECORDS. 33553—Bull. 61—08——8 102 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. region, but it is significant that the records in the plains region of Texas are all on rivers that flow eastward into the Gulf of Mexico, while the Kansas, Nebraska, and Iowa records are all in the vicinity of tributaries of the Mississippi. This does not prove conclusively that the form only occurs along streams in the grass-land regions, but since, I believe, it will generally be found closely associated with aquatic habitats even in the more humid and forested parts of its range, it would be surprising if it had a wider distribution in the more arid and treeless parts of the region in which it occurs. Variation.—Proximus exhibits such stability in most of its char- acters that it should be comparatively easy to detect the nature of such differences as do occur. The number of scale rows is inva- riably 19-17 throughout the greater part of its range. Indeed, I have observed but a single specimen with a different formula, a speci- men (No. 755) from Orizaba, Veracruz, in the U.S. National Museum, which has 17-19-17. It should be noted that this locality is toward the southern limit of the known range of the form. While the superior labials are somewhat more variable (fig. 36), the variations are still slight. The only variations observed are one Mt. Carmel, St. Louis, Ames, New Orleans, Kansas, Texas. Veracruz, Olney, Missouri. Iowa. Louisiana. Nebraska. Yucatan. Illinois. Fig. 36.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUPRALABIALS IN THAMNOPHIS SAURITUS PROXIMUS. specimen from Progreso, Yucatan, and two from New Orleans, with 7 on both sides, and one with 7-8 from Belair, Louisiana, one speci- men with 7 from St. Louis, Missouri, one with 7-8 and one with 7 from Olney, Illinois, and two with 7 and one with 7-8 from Mount Carmel, Illinois. These are slight variations, but it is significant that the tendency toward a decrease is only shown near the eastern and southern limits of the range. In the case of the number of dorsal scale rows and supralabials the variations are in all cases in the nature of a reduction. In the number of inferior labials, however, we find both an increase and decrease from the usual number, 10. The only two variations observed are 9 and 11, and these occur so rarely and so generally throughout the range that much larger series must be exam- ined before we can hope to detect any geographic differences in this character. I have stated that prorimus has from 150-179 ventral plates. It is with reluctance that I give the table below (fig. 37), for I have not enough specimens to plot the sexes in equal proportions and have VARIATIONS OF GARTER-SNAKES. 1038 had to base my averages upon all of the specimens available,and I am well aware that the sexual variation is sufficient to render the averages very deceptive in such small series. If care is taken, how- ever, not to attach too much importance to the averages, I believe that the table indicates very plainly that the number of ventral plates is quite constant over the North American portion of the range, and that from the State of Veracruz southward, in Mexico, there is a decrease in the number of scales in this series. This decrease in the southern part of the range is shown not only by averages from the different localities, but also by the fact that the smallest number of ventral plates observed in one hundred specimens examined from the United States is 158 (which occurs in but two specimens from New Orleans), while both specimens from Yucatan, in which the scales 150 Mt. Carmel, Ames, Douglas, New Orleans, Perry, Texas. Veracruz, Olney, Iowa. Kansas. Louisiana. Louisiana. Yucatan. Illinois. Fig. 37.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF VENTRAL SCUTES IN THAMNOPHIS SAURITUS PROXIMUS. have been counted, is as low as 150. Further than this the maximum number for six specimens from the State of Veracruz is 161, which is lower than any number observed in the United States, with the exception of the two New Orleans specimens. I readily grant that the series of specimens examined is too small to furnish exact evidence of the amount of decrease, but I believe that the records do indicate plainly that a reduction in the number of ventral plates actually takes: place in southern Mexico. There is no evidence, however, of a decrease in the number of plates in this series toward the eastern part of the range, comparable to that shown by the supralabials. As before mentioned the tail in proximus is comparatively long; the proportionate length observed varies from .25-.37. Although this variation is considerable, the extremes are seldom reached and the mean length of tail is about .29 or .30 throughout the range. 104 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. There is thus no geographical variation shown conclusively in the averages, but it should be noted that the longest tails are found in Louisiana, which may indicate a tendency toward an increase in the length of this organ in this region (fig. 38), while the shortest tails occur in Texas. Less evidence is at hand regarding the number of subcaudal scutes than is the case in any other series of scales, largely owing to the fact that the tail of many specimens is broken. The extremes are 75-125, an extent of variation which is equaled in no other form in the genus. The averages are not to be relied upon, but it may be seen from the table (fig. 39) that the smallest number occurs in the southwestern part of the range and the largest number in Louisiana. In this case, however, the variation in the number of subcaudal scutes is apparently associated with a variation in the tail length and affords no evidence of the dwarfing shown by the other characters. A| ce) Mt. Carmel, St. Louis, New Orleans, Tule Canon, , Austin, Veracruz, Olney, Missouri. Perry, Pecos, San Brownsville, Yucatan. Illinois. Louisiana. Angelo,Texas. Texas. Fig. 38.—DIAGRAM SHOWING THE VARIATION IN THE PROPORTIONATE TAIL LENGTH IN THAMNOPHIS SAURITUS PROXIMUS. There are few variations in coloration that deserve mention. Occa- sionally the dorsal stripe is red, but how frequently we do not know, since this color rapidly disappears in alcohol. Occasionally, also, this stripe becomes somewhat obscure, but both of these characters are apparently liable to crop out almost anywhere in the range. The type of Lutenia rutiloris Cope is characterized- by this obscurity of the dorsal stripe, although the latter is visible throughout the entire length of the body. The only other character (and the prin- cipal one) upon which this form was based was the presence of red upon the ‘‘superior and inferior labial plates and the first three large gastrosteges.”” This color is now faded and I have never observed a specimen of proximus which exhibited it, but the specimen differs in no other way from the Yucatan specimens of proximus, and the character is of too little importance to justify the separation of this specimen from proximus (See Ruthven, 1906a). To summarize the variations in proximus: — VARIATIONS OF GARTER-SNAKES. = “DOS (1) The dorsal scale formula is 19-17 throughout the range, except in southern Mexico, where a specimen with 17-19-17 has been examined. (2) The superior labials are 8 over most of the range, 7 only being observed in Yucatan and in general along the Mississippi River. (3) The inferior labials are usually 10, occasionally 9 or 11, but the variations are so evenly distributed over the range that no geo- eraphic differences can be detected in the small series examined. Mt. Carmel, St. Louis, New Orleans, TuleCanon, Austin, Veracruz, Olney, Missouri. Perry, Louis- Pecos, San Browns- Yucatan. Illinois. jana. Angelo, ville, ; Texas. Texas. Fig. 39.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUBCAUDAL SCUTES IN THAMNOPHIS SAURITUS PROXIMUS. (4) The variation in the number of ventral scales is large (150- 179), so that from the small amount of material available little reliance can be placed on the averages, but the number (both the averages, maximum and minimum) is strikingly smaller in Veracruz and Yuca- tan specimens than in the series from North America. (5) The tail length is very variable, although the mean length for the different localities is quite constant. Southwestern specimens 106 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. have the lowest number and Louisiana specimens the highest, which may indicate a tendency toward an increase in tail length in the latter region. (6) The number of subcaudals as it is correlated with the tail length is variable, so that the averages in the small series are not to be depended upon. But the largest number oceurs in Louisiana and the lowest in the southwestern part of the range in harmony with the variations in tail length. Owing to the correlation of the two, the variation in the latter obscures any tendencies toward dwarfing that may exist. (7) The color variations are too slight to permit of the formation of any geographic races on this basis. If, then, our material can be relied upon, there is indicated a tend- ency toward a decreased number of dorsal scale rows and ventral plates in southern Mexico and a decreased number of supralabials in general along the Mississippi River, which if true may be considered as an evidence of dwarfing in proximus in these two extremities of the range, while there is an apparent increase in the tail length in Louisiana, as shown by measurements and the number of subcaudal scutes. Affinities.—If we admit the evidence of the position of the lateral stripe, proximus is closely allied to sauritus, sackeni, and the members of the Radix group. Its relation to sauritus and sackeni is very close, as will be shown later; but its affinities with the Radix group are less clear. The chief character which separates it from this group is the decidedly longer tail; but, as shown above, the length of tail decreases in the southwestern part of the range, thus lessening the distinctive value of this character. It is in this region, therefore, that the form most closely resembles the forms of the Radix group, and it is here where the closest affinities must be sought. In northern Mexico and western Texas it differs markedly in scutellation from the representatives of the Radix group (megalops and marcianus) only in the smaller number of scale rows (19-17 instead of 21-19-17). This difference, as already seen, is only of racial importance, and as the formula is but one step less than that of the Radix group in this region, there can be no objections on this score to the direct devia- tion of proximus from marcianus or megalops. On the other hand, the range of proximus overlaps the range of marcianus to a consider- able extent without effecting the distinctiveness of either type, which argues against a direct relationship between the two forms, and the same is true of radix. Concerning megalops, however, the case is different. The ranges of the two forms come close together in Mexico, but do not overlap, megalops inhabiting the plateau and proximus the coastal plain. The tail length in prowimus is also less different from megalops than \ VARIATIONS OF GARTER-SNAKES. 107 from any other form in the Radix group, both owing to the somewhat greater length of tail in megalops and to the shorter tail in proximus in northern Mexico. Since the scutellation is practically the same in the two forms, except that the dorsal scale formula in proximus is slightly smaller, it may be that proximus is a direct relative of megalops which has pushed eastward into the eastern coastal-plain region of northern Mexico, and from here southward and northward into the regions which it now occupies. However, it is quite distinct from megalops at the present time, in the long tail and more slender body, and it is to the northward that we must look for its closest relatives. SACKENI.4 Description—This beautiful snake has received but little atten- tion from herpetologists and is represented in collections by but few specimens. The scutellation may be defined as follows: Dorsal scale rows, 19-17; supralabials, 8; infralabials, 10; ventral plates, 154-171; subcaudal scutes, 109-134; tail length, .32-.38. The col- oration is similar to light specimens of proximus, the lateral stripe being on the third and fourth rows (yellowish or greenish), the ground color usually a shade of brownish olive, and the belly and labials light and generally unmarked. As in proximus, the ventral spots are usually absent, but not always, and the lateral spots are generally fused irregularly, although occasionally distinct in the usual two rows between the stripes; even when distinct on the skin, however, they are not represented on the scales except by a narrow black border along the dorsal, and occasionally the lateral stripes. The lateral stripe is always distinct, but the dorsal is generally obscure and often entirely wanting; when present, it covers all or nearly all of the median three rows. Habits and habitat relations.—Sackeni is a very interesting snake. In the length of tail and slender body it represents the extreme in the group and genus, and with the attenuated form is coupled an agility of movement that we doubt is exceeded in any other garter- snake. While comparatively little is written upon its habits, there is slight doubt but that, like proximus, sackent is quite aquatic in its habits. Thus, Ditmars (1907, 219-220) writes, ‘In habits this species is very similar to the ribbon snake [sauritus]. It is very aquatic, and is seen frequently sunning on the branches of bushes that overhang the water, into which it drops when alarmed.”’ Loenn- berg (1894, 329) remarks of the specimens obtained by him in Florida, “All were caught in the grass in wet places.’’ Personally I have been able to make few observations on the habits of this form. During the summer of 1905 I reared two broods which were born on July 19. The young resembled those of other forms in habits except for a@ Thamnophis sauritus sackeni(KENNicoTT), Proc. Acad. Nat. Sci. Phila., 1859, p. 98. 108 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. their greater agility. Like the adults, they steadily refused earth- worms, but at the age of three days took to the water and captured and ate live minnows voraciously. Their manner of catching fish was interesting. Dropping or climbing into the basin they would rush about, mouth open, until they encountered a fish, when they would rush out of the water, lashing their tails energetically, carry the fish to a corner and proceed to devour it. The sense of sight seemed to be depended upon but little in capturing fish, and dead ones were eaten apparently as frequently as live ones. Range.—As at present known the range of sackeni is confined to the southern part of the coastal plain, in southern Mississippi and Florida. This physically recent feature with its low altitude (no- where more than a few hundred feet above sea level) is characterized by scores of stagnant rivers, lakes, lagoons, and swamps. The temperature and humidity are high and the rainfall-evaporation ratio exceeds 110 per cent (Transeau, 1905). The vegetation is rich, and consists of such forms as white cedar, sweet bay, magnolia, tupelo gum, swamp cottonwood, cypress, Quercus texana, etc., in the swamps, and several species of pines on the higher ground. Specimens of true sackeni have been examined from the following localities: St. Johns River, Volusia County, Georgiana, Palatka, Orange Hammock, Kissimee River, Kissimee, Enterprise, Lemon City, Little Sarasota Bay, Clear Water, Pensacola, Marion County, Gainesville, and Orlando, Florida, and Bay St. Louis, Mississippi. As far as I have been able to find, the form has never been recorded outside of Florida, although Ditmars (1907, 219) states that it is dis- tributed in the ‘‘coast regions of South Carolina and Georgia; Florida generally.’ Certainly typical sackeni may be expected to occur somewhat north of the latitude of the northern boundary of Florida, but in this general region it comes in contact with sauritus and the status of the two forms in the intermediate region must be examined before the northern boundary of sackeni can be even approximately fixed. I must confess to have examined but very few specimens from the debatable region, but the fact that sauritus specimens from the coastal plain from North Carolina northward show a much closer affinity to sackent than those from central Alabama would seem to indicate that true sackeni pushes farther up the Atlantic coast than in the interior, possibly into Georgia and South Carolina, as Ditmars indicates, which might also. be expected in view of its more aquatic habits and its association with the coastal plain con- ditions throughout the greater part of its range. The range as known at present is indicated on the accompanying map (fig. 40). Variation.—The dorsal scale formula is easily disposed of, for in every specimen examined it is 19-17. The labial formula is nearly VARIATIONS OF GARTER-SNAKES. 109 as constant, although the diagram (fig. 41) indicates that there is some variation in the number of superior labials. Thus, while 8 supralabials is the rule throughout most of the range, 7 occurs in the Fig. 4).—DISTRIBUTION OF THAMNOPHIS SAURITUS SACKENI, AS INDICATED BY THE LOCALITY RECORDS. western (southern Mississippi and western Florida) specimens, although in what proportion it is impossible to estimate, as very little material is available from this region. The inferior labials are 10 in Bay St. Louis, Pensacola, Gainesville, Kissimmee, Lemon City, Mississippi. Florida. Marion Co., Georgiana, Florida. Florida. Enterprise, Orlando, Little Sara- sota Bay, Florida. Fig. 41.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUPRALABIALS IN THAMNOPHIS SAURITUS SACKENI. every individual examined. The number of ventral plates is rather variable (fig. 42). The extremes are 154-171, the average number lying between 158-166. There is no definite geographic variation in 110 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. this trait indicated by the material examined. The tail length is strikingly constant for this group, which may be in part but not entirely explained by the small amount of material; the length ex- ceeds that of any other form in the genus, the extremes being .32—.38, the average between .33-.36 (see fig. 43). As is to be expected from 174 150 Bay St. Louis, Pensacola, Gainesville, Kissimmee, Lemon City, Mississippi. Florida. Marion Co., Georgiana, Florida. Florida. Enterprise, Orlando, Little Sara- sota Bay, Florida. Fig. 42.— DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF VENTRAL SCUTES IN THAMNOPHIS SAURITUS SACKENT. the length of the tail, the number of subcaudal scutes also exceeds that of any other form. As shown by the diagram (fig. 44), the extremes are 109-134 and the averages for the different localities 3ay St. Louis, Pensacola, Gainesville, Kissimmee, Lemon City, Mississippi. Florida. Marion Co., Georgiana, Florida. Florida. Enterprise, Orlando, Little Sara- sota Bay, Florida. Fic. 43.—DIAGRAM SHOWING THE VARIATION IN THE PROPORTIONATE TAIL LENGTH IN THAMNOPHIS SAURITUS SACKENT. between 120 and 127. Here again no geographic variations can be detected. As is generally the case in the forms of the Sauritus group, there is little variation in coloration in sackeni. The usual variation of light to dark in the ground color is present, the lighter individuals being a fs VARIATIONS OF GARTER-SNAKES. £10 light greenish olive, the darker ones often dark brown. The lateral stripe is always present, but, as stated above, the dorsal stripe is usually obscure. Still, although sackeni is usually described as hav- ing the dorsal stripe obscure or wanting, this is decidedly not always the case, for in some of the specimens in the U. S. National Museum from Georgiana, Florida, it is not only well defined but also of a bright color. This variation is not geographic, but may occur in individuals apparently anywhere in the range. Affinities. Fortunately there is little difficulty in determining the affinities of this form. The position of the lateral stripe on the third and fourth rows and the length of tail at once proclaims it a member 108 Ray St. Louis, Pensacola, Gainesville, Kissimmee, Lemon City, Mississippi. Florida. Marion Co., Georgiana, Florida. Florida. Enterprise, Orlando, Little Sara- sota Bay, Florida. Fic. 44.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUBCAUDAL SCUTES IN THAMNO- PHIS SAURITUS SACKENI. of the Sauritus group. It differs from proximus only in the greater average length of tail and the consequent larger number of subcaudal scutes, a smaller number of ventral plates, and the frequent obscurity of the dorsal stripe. A knowledge of the variations, however, shows that these differences are slight, for (1) the dorsal stripe in proximus is occasionally obscure, (2) the length of tail and number of sub- caudal scutes in proximus apparently increases, while the number of ventral plates decreases in southern Louisiana to approximate the condition in sackent. As a matter of fact there is no reason to believe that the two forms do not intergrade perfectly in southern Mississippi. It should be borne in mind, however, that although sackeni has a longer tail and more subcaudal scutes than proximus it apparently 112 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. tends to be more dwarfed, as shown by the reduced number of ventral plates and possibly by a tendency toward 7 instead of 8 superior labials. Its relation to sauritus will be considered later. SA URITUS.4 Description.—This is the best and longest known form in the eroup. The lateral stripe is upon the third and fourth rows through- out the length of body. The tailis longer than in proximus, but not as long as in sackent, the length varying from .29 to .36. The scutella- tion may be described as follows: 19-17 dorsal scale rows; 7, occa- sionally 6 or 8, superior labials; 10, occasionally 9, rarely 11, inferior labials; venteal plates, 150 to 172; subcaudal scutes, 87-137. The speta color above is usually chocolate brown, but varies from light olive brown to black. Michigan specimens are quite frequently black, with the exception of the keels of the scales, which are light boo. The stripes are nearly always bright and conspicuous, the dorsal generally having an orange tint, the laterals paler and of a greenish cast. The labials are usually without black blotches and the lateral spots are seldom distinct, although they are not always entirely fused to the total abolition of the hight (whitish) interspaces. As in proximus and sackeni, the ventral spots are generally absent. Habits and habitat relations.—Like the other members of the group, sauritus seems to be more than ordinarily aquatic in its habits, but apparently less so than either proximus or sacken. In Michigan we have generally found it about the margin of ponds and streams in damp woods. It is somewhat of a climber, and is occasionally found in bushes, several feet from the ground. When pursued it elides through the pools and herbage at an astonishing rate, and does not hesitate to take to water and conceal itself among the water plants, but it generally remains near the surface, and we have never observed it dive to the bottom like a natricid snake. There are numerous short notes in the literature, to the effect that sauritus prefers damp situations. Ditmars (1907, 217-219) states that it feeds on salamanders, tad- poles, frogs, and fish, but, like sackeni, refuses earthworms. Two other writers, Atkinson (1901, 151) and Surface (1906, 142-143), record insects in stomachs examined. The latter gives the follow- ing as making up the stomach contents of Pennsylvania specimens: Earthworms, spiders, insect fragments, ants, Plethodon cinereus, Spelerpes bilineatus, Hyla versicolor. It should be noted that the insects may have been contained in the stomachs of the frogs and salamanders. The number of young is comparatively small; we have counted the embryos in a few specimens, and they seem to aver- age about a dozen. a Thame sauritus (LINN 2 can Syst. Natur., XII, p. 385. VARIATIONS OF GARTER-SNAKES. pS; -Range.—As is well known, sauritus is found in the forest region of eastern United States. Characterized by mild temperatures and a plentiful rainfall, this region supports an abundant arboreal vegeta- tion that extends to the northward to the limit of tree growth and to the westward to the prairies. It includes two types of biota, the northeastern conifer, and southeastern deciduous forest types. The northeastern coniferous forest is characterized by such trees as the larch, balsam, white and black spruce, white and red pine, etc., and centers in the Laurentian highlands, while the southeastern deciduous forest type possesses the sugar maple, white ash, beech, and several oaks, and centers about the southern Appalachians. These two types of biota intergrade for a considerable distance in Michigan, New York, New Hampshire, and Vermont, although the higher parts of the Appalachians carry an outlying tongue of the northeastern coniferous type as far south as Georgia. The deciduous forest reaches its greatest development in southeastern United States and, as already noted, grades into the prairie through the so-called ‘fringe forest,’ which covers all of Ohio and Indiana, and margins the prairie in Hlinois, Missouri, southeastern Kansas, and east-central Texas. Specimens of sauritus have been examined from the following localities: Roscommon County, Alma, Olivet, Washtenaw County, Lapeer County, Rawson Lake, and Lansing, Michigan; Waterloo, Lake Maxinkuckee, Veedersburgh, Lebanon, and Wheatland, Indi- ana; Medina County, Toledo, and Oberlin, Ohio; Delaware County, Darby, Londongrove, Allegheny County, Carlisle, and Indiana County, Pennsylvania; Haddonfield and Pleasant Point, New Jersey; Tioga County and Highland Falls, New York; Middletown, Connecticut; Wellesley, Sherborn, Woods Hole, and Lancaster, Massachusetts; Auburn, Maine; Chepachet Island, Rhode Island; Arlington, Vir- ginia; Laurel, Maryland; Washington, District of Columbia; Avoca, Summerville, and Wilmington, North Carolina. These localities represent the entire range of the form, as at present known, and it is not necessary to cite the numerous records in the literature. From these records it is evident that, if the form is properly defined, the range is closely confined to the southeastern forest region, north of Florida and east of the Mississippi River and the prairie peninsula. The northernmost localities known to me are Roscommon County, Michigan, Norway and Auburn, Maine, which would indicate, as is undoubtedly the case, that the form occurs in extreme southern Canada. It will thus be seen that sawritus does not enter the northeastern forest proper, but pushes well into the inter- mediate region between this forest and that of southeastern North America, 114 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. | The western and southern limits are not definitely known. This — - is due in large part to the want of specimens from the regionpf these boundaries. As will be shown later, however, the western limit may probably be given in general as the Indiana-Illmois boundary and western Kentucky, Tennessee, and Mississippi (fig. 45). Variation.—From the list of localities from which specimens of sauritus have been examined it might seem that a considerable amount | Fig. 45,— DISTRIBUTION OF THAMNOPHIS SAURITUS, AS INDICATED BY THE LOCALITY RECORDS. of material was available for the study of this form. Unfortunately, however, none of these localities are represented by more than about a dozen specimens, the usual number being one to four. Even after combining the records from neighboring localities, the data is still insufficient to denote more than general conditions. As in the case of sackeni, the dorsal scale formula in no specimen examined departs from 19-17.4 The labial formula, on the other a Compare p. 39, VARIATIONS OF GARTER-SNAKES. 115 hand, is more variable than in either of the other forms in the group. An examination of the diagrams (figs. 46-47) will show that the variation in both series consists of a loss or an addition of one scute from the more constant number 7/10. The average number of supralabials is generally a little above 7, showing the comparative Wheat- Maxinkuc- Oberlin, Washte- Carlisle, Avoca, Washing- Delaware Rhode land, kee, Medina naw, Pennsyl- North ton,D.C., County, Island, Veeders- Waterloo, County, Lapeer, vania. Carolina. Laurel, Pennsyl- Massa- burg, Indiana. Ohio. Eaton, Mary- vania. chusetts, Indiana. and land. Pleasant Connecti- Gratiot Point, cut. counties, Haddon- Michigan. field, New Jersey. Fig. 46.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUPRALABIALS IN THAMNOPHIS SAURITUS. rareness of specimens with 8 or 6 scutes in this series. No definite geographic differences can be discovered in the material examined. In the case of the infralabials the conditions are exactly reversed in that, while the variations are so slight as to disturb the averages 17 Wheat- Maxinkuc- Oberlin, Washte- Carlisle, Avoca, Washing- Delaware Rhode land, kee, Medina naw, Pennsyl- North ton,D.C., County, Island, Veeders- Waterloo, County, Lapeer, vania. Carolina. Laurel, Pennsyl- Massa- burg, Indiana. Ohio. Eaton, Mary- vania. chusetts, Indiana. and land. Pleasant Connecti- Gratiot Point, cut. counties, Haddon- Michigan. field, New Jersey. Fig. 47.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF INFRALABIALS IN THAMNOPHIS SAURITUS. but little, the differences that do occur are mostly by way of a reduction to 9, so that the averages run a little below 10. As in the other ribbon snakes, the variation in the number of ventral plates is considerable. In the material examined the ex- tremes are 150 and 172. The table (fig. 48) will show how the 116 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. variations are distributed. While this diagram is very subject to error, owing to the small number of specimens, it is interesting to note that specimens from eastern localities have apparently a smaller number of ventral scutes than those from Michigan, Ohio, and Indiana, and while larger suites of specimens will undoubtedly 1714 166 158 150 Wheat- Maxinkuc- Oberlin, Washte- Carlisle, Avoca, Washing- Delaware Rhode land, kee, Medina naw, Pennsyl- North ton,D.C., County, Island, Veeders- Waterloo, County, Lapeer, vania. Carolina. Laurel, Pennsyl- Massa- burg, Indiana. Ohio. Eaton, { Mary- vania. chusetts, Indiana. and land. Pleasant Connecti- Gratiot Point, cut. counties, Haddon- Michigan. field, New Jersey. Fig. 48.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF VENTRAL SCUTES IN THAMNOPHIS SAURITUS. change the averages, we believe that the difference is a real one. The length of tail (fig. 49) is apparently quite constant. The ex- tremes are .29 to .36, the average about .33, and while the average can not be relied upon, it may be significant that the tail length is “| o2o Wheat- Maxinkuc- Oberlin, Washte- Carlisle, Avoca, Washing- Delaware Rhode land, kee, Medina naw, Pennsyl- North ton,D.C., County, Island, Veeders- Waterloo, County, Lapeer, vania. Carolina. Laurel, Pennsyl- Massa- burg, Indiana. Ohio. Eaton, Mary- vania. chusetts, Indiana. and land. Pleasant Connecti- Gratiot Point, cut. counties, Haddon- Michigan. field, New ; Jersey. Fig. 49.—DIAGRAM SHOWING THE VARIATION IN THE PROPORTIONATE TAIL LENGTH IN THAMNOPHIS SAURITUS. somewhat greater in eastern localities. This is shown better in the number of subcaudal plates (fig. 50). Thus, in Ohio, Michigan, and Indiana the extreme number of subcaudal plates is 87 and 121, the averages between 100 and 117, while along the coastal plain the extremes are 103 and 137, the averages between 115 and 120. (No ee ee eee VARIATIONS OF GARTER-SNAKES. Lay differences in color have been observed which are evidently geo- graphic.) If the data plotted can be depended upon to reveal the actual state of variation, the form may thus be divided into two sections upon the basis of the scutellation and tail length, i. e., a trans- Allegheny section characterized by a larger number of ventral scutes, a shorter tail and fewer subcaudal scutes, and a coastal plain section characterized by a fewer number of ventral plates, a longer tail and more subcaudal scutes. Wheat- Maxin- Oberlin, Washte- Carlisle, Avoca, Washing- Delaware Rhode land, kuckee, Medina naw, Pennsyl- North ton,D.C., County, Island, Veeders- Indiana. County, Lapeer, vania. Carolina. Laurel, Pennsyl- Massa- burg, Ohio. Eaton, Mary- vania. chusetts, Indiana. and land. Pleasant Connecti- Gratiot Point, cut. counties, Haddon- Michigan. field, New Jersey. Fig. 50.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUBCAUDAL SCUTES IN THAMNOPHIS SAURITUS. Affinities—The position of the lateral stripe (on the third and -fourth rows) and the long tail justifies us in placing sawritus in the same group with sackeni and proximus. As has been shown, it meets the range of sackeni in the latitude of northern Florida and that of prozimus along the eastern boundary of Illinois and in western Ken- tucky, Tennessee, and Mississippi. Specimens from the intermediate region along the common boundaries of the three forms remain to be examined, but I believe that there can be no doubt that sauritus intergrades perfectly with both sackent and proximus. A 2 ee ee a; oO 118 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. Sauritus differs from sackeni principally in its generally fewer number of labials, subcaudals, and ventral scutes and the shorter tail, and from proximus in the fewer labials and ventral scutes and greater number of subcaudals and slightly longer tail. A comparison of the tables, however, will show that these differences are not sharp. Thus, while the usual number of labials in sackeni are 8/10, the formula 7/10 is not rare, while in sauritus the general formula 7/10 is not infre- quently replaced by 8/10. This suggests that the two forms may intergrade in this trait, although the data are not sufficient to establish this geographically. It has previously been noted that proximus specimens from the eastern part of the range show an apparent tendency toward a reduc- tion in the number of supralabials from 8 to 7, the average number for Illinois specimens being almost identical with the average for Indiana specimens of sauritus. Very similar conditions exist in the sase of the other characters. In coastal plain specimens of sauritus the extremes of variation in the number of ventral scutes are quite close to those of sackeni, although the averages are somewhat lower. In southern Michigan, Ohio, and Indiana, however, the number of scutes in this series is apparently higher than in either the coastal plain specimens of sauritus or sackeni, which brings the form in this region so close to proximus specimens from Illinois, eastern Iowa, and St. Louis that the character is useless as a distinguishing one: Again, the greater number of subcaudal scutes and the tail length in coastal plain specimens of sauritus approximate the conditions in sackeni, while the shorter tail and fewer subcaudals in western specimens of sauritus are almost identical with eastern specimens of proxvmus. It seems very evident from this digest of the variations in the three forms (1) that the form sauritus intergrades both with sackeniand proxi- mus; (2) that the longer tail, more numerous subcaudals, and fewer ven- trals relate the specimens of sauritus from the eastern coastal plain more closely to sackeni, while the shorter tail, fewer subcaudals, and greater number of ventral scutes relate the interior specimens directly to proximus. I suggest, then, that the form now known as sauritus has had a double origin, 1. e., the coastal plain section from sackeni and the trans- Allegheny ‘section directly from proximus. This is what we might expect, for the fact that the ribbon snakes are so eminently adapted to the coastal plain conditions would favor the rapid invasion of the eastern coastal plain from Florida, while the Appalachiansystem would prove more of a barrier to such an aquatic form than to a more ter- restrial one, and permit the section from beyond the Mississippi to push into the region west of the Appalachians. However this may be, ee VARIATIONS OF GARTER-SNAKES. 119 it should be noted that sawritus as a form hag, exclusive of the dorsal scale rows, a smaller scutellation generally than either sackeni or proximus. CONCLUSION. In the foregoing pages we have endeavored to give the results of a study of the characters, habitat relations, variations, and affinities of the three forms, proximus, sackeni, and sauritus, which are apparently united into a single group by the position of the lateral stripe upon the third and fourth rows and the long tail. The evidence seems to indicate that the three forms intergrade perfectly with each other, but even though the evidence may be considered too slight to prove this, I believe it to be indisputable that the closest relationships of these forms are along the common boundaries (fig. 51). sauritus sackeni proximus Fig. 51.—PHYLOGENETIC DEVELOPMENT OF THE SAURITUS GROUP. Of the three forms, proximus, in northern Mexico and Texas, pos- sesses the maximum scutellation (exclusive of the subcaudal plates) and sauritus the minimum, sackeni possessing an intermediate number. The decrease takes place gradually from proximus to sackeni to the coastal plain section of sauritus and from proximus directly to the trans-Allegheny section of sauritus. The tail length and number of subcaudal plates increase somewhat from proximus to sauritus and markedly from proximus to sackeni, apparently decreasing again in the coastal plain specimens of sauritus, thus approaching the shorter- tailed conditions in the trans-Allegheny section of the latter. It must also be stated that according to my observations proximus specimens attain to the largest size. The center of origin of this group will be discussed later (see p. 187), but it may be stated here that I believe the center of dispersal to be in northern Mexico. If the group did originate in this region the explanation for its present distribution would seem clear. Being rather aquatic in its habits, we would expect the original stock to 120 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. adhere quite closely to the coastal plain in Mexico, but in the United States it has apparently been able to push out along the rivers, the westward trend of the streams permitting it to cross the plains and prairies, the Mississippi to extend its range well to the northward. East of the Mississippi River the group has been able to extend over all of eastern United States, the coastal plain leading one section from the coastal plain of Texas and Louisiana into Florida and from here up the Atlantic coast, the other section pushing into the trans- Allegheny States from the Mississippi Valley. It should be noted again in this connection that a reduction in scutellation and size in the group takes place away from northern Mexico, both to the south and east (the extremities of the range of the group), and that the area between the forms, in which a difference in scale formulas is accomplished, is in every case narrow. THE ELEGANS GROUP.—(ANGUSTIROSTRIS, MELANOGASTER, SCALARIS, PHENAX, HAMMONDI, ELEGANS, ORDINOIDES.) ANGUS TIROS TRIS .4 Description.—If in assembling these forms of garter-snakes I have distinguished a natural group I should be able to support my position by pointing out the lines of relationship between the com- ponent forms. As in the Radix group, I begin with the forms on the Mexican table-land, one of which (angustirostris) also possesses the maximum scutellation for the group. Apparently not over a dozen specimens of this form have been recorded, although it is probably not rare in the region where it is found. In appearance it is one of the’ most peculiar forms in the genus. The stripes are usually entirely wanting, but the laterals are occa- sionally faintly in evidence on the second or second and third rows, and the dorsal for a short distance anteriorly on the median row. The scutellation in the specimens examined is as follows: Dorsal scale formula, 21-19-17, with occasional variations to 21—23-21-19-17 and 19-21-19-17; supralabials, 8 to 9; infralabials, 10-11 (9 in one instance) ; preoculars, usually 3; postoculars, 3 to 4; subcaudals, 65 to 83; ventrals, 153-163; tail length, .235 to .256. The eye is generally cut off from all but one labial by the anterior prolongation of the lower postocular, but this is not always the case. The color above is usually dark olive brown, with six rows of dark brown spots on the scales, distinct in the young, but indistinct or obsolete in the adults. The head is also unicolored above in the adults, while in the young it is finely speckled with lighter. Each upper and lower labial is margined with a triangular patch of brown a Thamnophis angustirostris (KeNnNicorr), Proc. Acad. Nat. Sci. Phila., 1860, p. 332. This form includes Atomarchus multimaculatus Corr and Chilopoma rufopunctatum Cope. VARIATIONS OF GARTER-SNAKES. Lot bordered with darker. Belly greenish or bluish ash, marked with numerous small black spots, and a bar at the base of each ventral, which is usually prolonged outward to form a more or less irregular median ventral band. Throat yellow. Habits and habitat relations.—Little data is available on the nabitat relations of this snake, but it is undoubtedly particularly aquatic in its habits. Thus, Cope (1883b) states of the San Francisco River specimen obtained by him: The only specimen of this species which I have seen living was taken in a seine net with which I was fishing near the bank of the San Francisco River. It dove into the net, seeking the bottom of the water as a place of concealment, as is the habit of E. melanogaster and contrary to that of E. saurita and E. macrostemma, which swim preferably on the surface, seeking concealment under banks. Range.—As at present known, this form is confined to the northern partof the Mexican plateau (above northern Durango) and the propla- teau region in southwestern New Mexico and southeastern Arizona (fig. 52). The conditions of this habitat have already been described (p. 46). Specimens have been examined from Chihuahua and San Andreas, Chihuahua; Parras, Coahuila, and Coyotes, Durango, but it occurs also in southern Arizona, as Cope (1883b, 1300-1301) records a specimen from the San Francisco River in southern New Mexico, and the type of rufopunctatum came from ‘‘southern Arizona.”’ Variation.—Very few facts are available for the study of variation in this form. The dorsal scale rows are very constantly 21-19-17. A single specimen from San Andreas, Chihuahua, has 21—23-21-19-17, and another from Chihuahua, Chihuahua, has 19-21-19-17. No geographic variation is discernible in this character in the material at hand, the more northern and southern specimens all having gen- erally 21-19-17 rows. The same thing is true of the labial scutes, the formula being 8/10 or 8/11, or 9/10 or 9/11, but with no certain differences in those from different localities, although only those from Chihuahua have 9 in the superior series. In regard to the ocular scales the conditions are somewhat differ- ent. In all of the specimens examined from the vicinity of Chihuahua, Chihuahua, there are three preoculars and three or four postoculars, and the eye is excluded from the posterior subocular labial by the lower postocular, but in those from northern Durango and southern Coahuila the formula is 2-3 or 1-3 and the eye is fairly in contact with two labials. Similarly the type of rufopunctatum from “southern Arizona” is said to have had one preocular on one side and two on the other, as well as three and four postoculars, with the eye in con- tact with two labials. But little idea can be obtained as to the range or mean in the variations of the ventral and subcaudal scutes. In the material examined the males have in general 162—166 ventrals and 73-83 subcaudals, the females 153-159 ventrals and 64-69 sub- 129 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. eval =o} Be ~~~1In j-----r-- 7h M3 VW. we] FIG. 52.— DISTRIBUTION OF THAMNOPHIS ANGUSTIROSTRIS AS INDICATED BY THE LOCALITY RECORDS. — ~n ——- oe VARIATIONS OF GARTER-SNAKES. 123 caudals. The tail length varies from .22—.256, the average of males and females combined being about .225. It is needless to point out that these figures can not be relied upon to furnish more than a general idea of the scutellation and propor- tions in this form. Still, the more constant characters seem to indi- cate (1) that in the vicinity of the city of Chihuahua the number of dorsal scale rows, oculars, and possibly the labials reach the maxi- mui for the form, and also for the group and genus; (2) that toward the northern and southern limits of the range the number of oculars, especially the preoculars, tends to decrease; (3) that the orbit, which in every specimen from the vicinity of Chihuahua, Chihuahua, is separated from all but one labial by the inferior postoculars, becomes fairly in contact with two in the northern? and southern parts of the range. Affinities —In the most southern localities from which it is known angustirostris comes in contact with melanogaster. These forms always resemble each other closely in that both are characterized by a slender head, small eye, mostly 8 supralabials and 10 infralabials, nearly the same number of ventral scutes, more than one preocular, the absence of the dorsal stripe, and the presence of a dark median ventral band. The differences between them are that in specimens of angustirostris there are generally well-defined blotches on the labials, the lateral spots are usually in evidence, the lateral stripes generally obsolete, the eye usually in contact with but one labial, and not less than 21-19-17 scale rows. While in melanogaster there are no large blotches on the labials or evidence of lateral spots, the lateral stripe is often present, the eye is in contact with two labials, and the dorsal scale formula rarely exceeds 19-17. It may be seen from this that the differences between the two forms are slight except in two particulars, the segregation of the eye from the second subocular labial and in the dorsal scale formula. In regard to the former trait we have already shown that in the region which these forms hold in common angustirostris becomes exactly like melanogaster in the fact that the eye is in contact with two labials. It will also be shown in the consideration of melanogaster that the number of rows of dorsal scales tends to become as in angustirostris (21-19-17 or 19-21-19-17). The evidence, therefore, favors strongly the view that the two forms intergrade in northern Durango and southern Coahuila, so that it must be adopted for the present, although the relationships of the two forms in the intermediate region should aIn the type of multimaculata (from the San Francisco River) the eye is in contact with but one labial, but in rufopunctatum, as already mentioned, it is bounded by two. 124 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. be carefully investigated before any definite conclusion can be reached.¢ Elsewhere in its range angustirostris is not known to meet any similar form, but to the northward the known limit of its range approaches that of one which is strikingly like it in many respects. I speak of 7. hammondi (Kennicott). The poimts of similarity be- tween these forms will be discussed when hammondi is considered in detail. MELANOGASTER. Description.—l have already stated that this species resembles angustirostris so closely as to suggest strongly a near relationship. The head is slender and the eye small. Lateral stripes usually pres- ent, and confined to the second row of scales. The dorsal stripe is usually wanting, but may be present on the median row. Dorsal scale formula, 19-17; the labials, 8/10 (very rarely 7/10, 8/9, or 8/11); oculars, usually 2—2 or 2-3, occasionally 2—4 or 3-3; ventral plates, 139-158; subeaudals, 49-76; proportionate tail length, .19-.26. The color above is uniformly very dark brown without distinct spots. Both the dorsal and lateral stripes are indistinct when present. Belly greenish slate or yellowish ash, with a black bar on the base of each ventral, which is prolonged in the middle to form a median ventral band. Throat and supralabials dark yellow. Habits and habitat relations.—Like its northern relative, angus- tirostris, melanogaster is apparently quite aquatic in habits. The following account of its habits at the City of Mexico, recorded by Cope, has already been quoted in the discussion of the habits of megalops. aKennicott’s type of angustirostris (Parras, Coahuila) is not exactly typical of the northern form, as the eye is fairly in contact with two labials; but while it is thus some- what intermediate between the northern and southern forms, it is much closer to the former, as shown by the coloration and the possession of 21-19-17 scale rows, thus making it entirely fair to give this name to the form. bThamnophis angustirostris melanogaster (PETERS), Monats. Kénig. Preuss. Akad. Wissen. Berlin, 1864 (1865), pp. 389-390. Includes Tropidonotus baronis-miilleri TroscHeEL, Tropidonotus mesomelanus JAN, and Regina mesomelaena Cope. The fact that Peters states that ‘‘Diese Art ist bereits von Wiegmann (Preisverzeichniss der Stiugethiere, Vogel, Amphibien, Fische, und Krebse, welche von Hrn. Deppe und Schiede in Mexico gesammelt worden, Berlin, 1. Sept., 1830; wieder abgedruckt in Cabanis Journal fur Ornithologie, 1863, p. 54) benannt und unter dem obigen Namen versandt und verkauft worden’”’ has led Cope to credit this name to Wiegmann. Inasmuch, how- ever, as Peters first described the species, he should be held responsible for the name, for Article 21 of the International Code distinctly says that ‘‘ The author of a scientific name is that person who first publishes the name in connection with an indication, a definition, or a description, unless it is clear from the contents of the publication that some other person is responsible for said name and its indication, definition, or description,’ and Wiegmann, as is well known, gave no description of this species. | j VARIATIONS OF GARTER-SNAKES. 125 The E. macrostemma is the more active, sooner seeking the water, where it swims keeping close to the shore, and remaining more or less in sight until it conceals itself inahole. ‘The LZ. melanogaster, on the other hand, lies quietly so as to be more easily taken in the hand; but, if it once takes to the water, it seeks the depths and is no more seen. It is much less disposed to bite than the EL. macrostemma; the latter being, like its ally, the L. sirtalis, a very pugnacious snake. A number of specimens have been taken about Lakes Xochimilco and Chaleo, Mexico. Range.—In contrast to angustirostris, which occurs on the northern part of the Mexican plateau, melanogaster has thus far been recorded only from that part of the plateau which lies south of the 26th par- allel. From this latitude it extends southward to Puebla, but has not been recorded from that State. Over the northern part of this region the environmental conditions are the same as in the northern part of the plateau, but to the southward the precipitation mcreases with the result that permanent rivers and lakes are formed. Specimens have been examined from the following localities: Coyotes and Durango, Durango; Ocatlan, Jalisco; Lakes Xochimilco and Chaleo, and Mexico City, Mexico; besides a number of others with no other locality label than ‘‘Mexico.”’ Giinther (1894, 134) records the following general localities: Guanajuato, Veracruz, Tehauntepec. Owing to its relationship to angustirostris, we do not believe that it occurs north of Coyotes, Durango, but the locality records probably do not represent the southern limit of its range tiie} 53). Variation.—As in the case of angustirostris, far too few specimens of this snake are available to reveal whether or not geographic vari- ation in scutellation exists. Certain general facts should be pointed out, however. The dorsal scale formula is 19-17 in every specimen examined, the ventral plates vary in number from 139-158 (average 145-150), the subcaudals from 49-67 (average about 58). These numbers indicate plainly that melanogaster has a distinctly reduced scutellation from the maximum for the genus, a fact which is not disproved by the labial formula. Thus, the supralabials are nearly always 8, the only differences consisting of a loss of a single plate, while the normal number of infralabials (10) becomes occasionally 9, and in but one instance 11. The tail length, as previously stated, varies from .19 (female) to .26 (male), the average length for every locality with two or more specimens being .22. Affinities —I have already pointed out in the discussion of the affinities of angustirostris that these two forms are evidently related in that the strongest distinguishing characters possessed by angus- tirostris (the segregation of the eye from all but one labial and the presence of more than two preoculars) apparently disappear in that part of the range which it shares with melanogaster. It is natural 126 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. to inquire now what particular evidence, if any, melanogaster shows of such consanguinity. Since we are inhibited by a lack of speci- mens from determining geographic variation, we can expect but little evidence from one form when the two are so very similar. Fig. 53.—DISTRIBUTION OF THAMNOPHIS ANGUSTIROSTRIS MELANOGASTER, AS INDICATED BY THE LOCALITY RECORDS. Specimens of the two forms are so nearly alike both in scutellation and color, that.-it is impossible to give any character that will always serve to distinguish them; but whether or not we grant that inter- gradation occurs at the present time, I hold it to be quite evident VARIATIONS OF GARTER-SNAKES. 1 ir that. they are directly related. In this connection it should be noted that apparently in every series of scales melanogaster is more reduced than angustirostris. In the south melanogaster comes in con- tact with two very imperfectly known forms with which I believe it also to be related; these are scalaris and phenax, which will next be discussed. SCALARIS.4 Deseription.—lIt is nearly always difficult to properly determine the characters and affinities of forms which are known from but few locali- ties, owing to the fact that geographic differences are seldom revealed. This is particularly true of the garter-snakes of southern Mexico and Central America, a region from which a number of species have been described, none of which are represented in American museums by more than a few specimens. - Of these forms scalaris is perhaps the best known, but, although access has been had to the material in the U.S. National Museum, Academy of Natural Sciences of Philadelphia, and the Field Museum of Natural History, it has been possible to find a total of but nineteen specimens, representing five localities. As known at present, the form may be described as follows: Lateral stripe on the second and usually the adjacent edge or half of the third rows. Eye small. Dorsal stripe principally confined to the median row. Dorsal scale formula, 19-17, 19-17-15, 17-19-17, 17-19-17-15, or 17-15; supralabials, 7, very rarely 8; infralabials, usually 9, occa- sionally 8 or 10; oculars, 1-3 or 1-2; ventrals, 134 (female) to 151 (male); Subcaudals, 52 (female) to 76 (male). Ground color above brownish ash, olive brown, or dark chocolate brown; the first row somewhat lighter. Stripes light ash to dull orange yellow. In some specimens there are no lateral spots to be observed on the scales, and in others there is a row of dark transverse bars between the third and seventh rows, inclusive, which alternate with a series of small spots on the eighth and ninth rows. In the ma- jority of specimens, however, there is on either side a single series of large (one and one-half scales wide and one apart) rectangular, trans- verse, dark brown spots with dark edges, which extend from about the middle of the third to the median row of scales. These spots are well defined from the nape (where they are represented by a pair of large nuchal blotches) to the anus, are as distinct in the young of the single brood examined as in the anletey and the spots of ach side are mle separated by the dorsal stripe on the medianrow. Head above brown, generally marked with dusky yellow. Postoculars and supralabials dusky yellow. The plates in the latter series may not be margined with darker. a eae ae Gorn, Beoe Acad. Nat. Sci! Priiiay 1886, Di 306: Includes T. scaliger JAN. 128 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. Habits and habitat relations.—N othing is known of the habits of this form, but it is significant that several specimens in the U.S. National Museum, collected by F’. Sumichrast, are labeled ‘Alpine region,’ by which is probably meant the pine zone surrounding the zone of grasses and herbaceous forms which margins the snow fields and glaciers of the summit. Dr. W. L. Tower informs me that the pine zone attains a maximum elevation of 13,000 or 13,500 feet. That scalaris occurs in this habitat is shown quite clearly by the fact that a series of speci- mens in the Field Museum of Natural History were taken at an eleva- tion of 13,000 feet. These are the only available notes on the habitat of the form. Range.—All that is known of the range of scalaris is indicated by the following records: Guanajuata; Jalapa and Mount Orizaba, Vera- Fig. 54.—DISTRIBUTION OF THAMNOPHIS SCALARIS, AS INDICATED BY THE LOCALITY RECORDS. cruz; Valley of Mexico and Toluca, Mexico; Tecpam, Guatemala; Talpa and Autlan, Jalisco; Oaxaco. We have examined specimens from all but the last three localities (Boulenger, 1893, 204). (Fig. 54.) Whether or not it is of general distribution in the region indicated by these records can not be conjectured, as practically nothingis known of its habitat relations. Variation.—It is highly inexpedient to attempt to draw conclusions on the nature of the variations in such a small series as is here available more especially since fifteen out of the nineteen specimens hail from one locality. In the accompanying table, however, we give the scutel- lation and measurements of all of the specimens examined, and not- withstanding the scarcity of material certain facts stand out quite prominently. Thus, the sixteen specimens from Orizaba and Jalapa, VARIATIONS OF GARTER-SNAKES. 129 Veracruz, all have 17-19-17 scale rows or less, while none of the three specimens from the States of Mexico and Guanajuato have other than 19-17-15, and one of the Guatemala specimens has 19-21-19-17. Such a reduction in the number ot scale rows would represent a dwarf- jng within the form, but, regardless as to whether or not this apparent tendency is borne out by future investigations, it is quite evident from the predominance of the smaller dorsal scale formulas together with the evidently small number of labials that the form is, as a whole, a dwarfed one, so much so indeed that the number of scales in the dorsal, supralabial, infralabial, and ventral series attains the minimum for the genus, the form being more reduced than butlert in this regard. It should also be noted that the size of all of the specimens obtained is small. Measurements and scutellation of Thamnophis scalaris Cope.4 | ca 2 a » | a : te Museum. pea Locality. Dorsals. - ‘3 2 a|= |) ie & = ; = i ehe lie aio iss Blf/s/8/a|/8 |s lesa =] q i) 3 o lo} a lors Wiles | S| wm | & BSB |e | z: 2 : — lrg | mm \mm} Field Museum..... 1517 | Orizaba, Veracruz. ...| 17-19-17-15 | 7 fe ae GI Was es eee | | See 18 ee Dee 1517 | Mount Orizaba,Vera-| 17-15 =| 7| 9 | 1-2| 74| 138| 172 | 47| 0.27 cruz. | | DOR. 0 pce WONT | eae er Ome cers tvetee eee. 17-15 7 eid Easel Cal BY essa laarel eenees 9 ID Oma Sten aes OAS Ssce CONO}e I ae ia eee 8 17-15 7 re he 65 | 136 | 173 | 44 25 1D is Sepa seee LoMa asec GOBee sek es 17-19-17-15 | 7 Oh S-8h- | 66 | 140 | 175 | 438} .24 | 1-2 | Big eee GRU Ale sees Gp a eae 17-19-17-15 |7| 8 if : loo 138 | 178 | 42| .23 IDES See re IGS U7 (0) (ees (OKO Ae ee Aes Oe 8 17-15 7 8 | 1-2 | 75 | 137 | 170 | 47 27 8 1D YO}e See eae parents GNF fal Eee GOA Se ee ee 17-15 U { 9 {2 75 | 138 | 180 | 50 20 DORE Naess 15230 eae (ONO Sena Pre, EE Ee 17-19-17 7 9 | 1-3 | (ey Cea Varo I ae U.S. National Mu- | 12115 | Orizaba, Veracruz....| 17-19-17-15 | 7 | 9 | 1-3) 59 | 135 | 347 | 82 23 seum. | [ne DOs 5 eae 30497 |...-.. GOs saa epee eee 17-19-17-15 | 7 | 10 | 1-3 | 75 | 145 | 316 | 81 20 | “Alpine region,” Ori- || 8 | 7 17-19-17-15 +10 | 1-3 | 58 406 22 Do........---- | 7076 || zaba, Veracruz. | "17 | = ial 8 DORs ane ee ZO7O8| oe ee COS oe = See eee 17-19-17-15 | 7 9 [is 76 | 134 | 185 | 50 27 DOR see a. (CUPS) eeeee Gla)? See BAA Seee Cee 17-19-17-15 | 7 9 | 1-3 | 72 | 187 | 207 | 53 | 25 af 1-2 DOreracee secs he CA See CO eee ees eae ea 17-15 | ti 3 {rs 142 | 335 |100 29 Do.........-...|-32281 | Valley of Mexico and 19-17 7| 91} 1-3 | 52 | 143 | 385 | 79 -20 Toluea. 1D) Oeeee oes 12675 | Guanajuato..........- 19-17-15 7 | 9 | 1-2 | 56 | 144 | 387 | 80 20 | =9) DOssse sseesce2 | 12730 | Mexico, Mexico.....-.. 19-17 fiM 2) {i z |s2 151 | 410 | 78 -19 | . Academy Natural) 11694 | Jalapa, Veracruz. ... -| 17-19-17-15 | 7 | 9 | 1-3 | 73 | 138 | 340 | 95 28 Science, Phila- delphia. | | | | | _ 7Since this table was compiled I have examined four specimens of scalaris from Tecpam, Guatemala, in the Field Museum. Three have 19-17 and one 19-21-19-17 scale rows. 130 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. Affinities—Little reliance can be placed on efforts to determine in more than a general way the affinities of this form, but something can be done in this direction. In the first place, if we can pin our faith to the constancy of the trait, the position of the lateral stripe on the second and third rows at once excludes scalaris from the Radiz and Sauritus groups. On the basis of our classification, then, it is to be referred either to the Sirtalis or Elegans groups. The latter has otherwise, so far as we know, no representatives in this region, but the ground for referring it to this group is mostly the negative one that the former group (Sirtalis) covers the same territory as_ scalaris, which argues against a relationship in this direction. If scalaris is a derivative of the Elegans group it is quite important that its relationship with melanogaster (the nearest form, geographic- ally, of the same group) be determined by careful collecting and observation. From the present data the ranges of the two forms seem to overlap in southern Mexico, but the habitats of the two may be quite distinct, a conclusion that is strengthened by the fact that on Mount Orizaba scalaris is found, to some extent at least, at a considerable altitude. On the other hand, it is not impossible, and I rather incline toward the hypothesis, that scalarzs is more closely related to elegans than to any other form, and represents an offshoot of the latter which has pushed southward from the range of elegans, in the mountains of eastern Mexico. In defense of this view, it is interesting to note that elegans prefers comparatively high altitudes in the Rocky Mountains. At any rate, it is, as known at present, a well-defined form, which is liable to confusion only with phenaz. Its relations to phenax will be discussed in considering the affinities of the latter. PHENAX.@ Description.—The status of this form is much the same as that of scalaris. At the present time I know of but two specimens in Ameri- can museums. Cope writes that specimens have been sent to both the Academy of Natural Sciences of Philadelphia and the U. 5S. National Museum, and records six specimens from the latter mstitu- tion, but the only ones that I have been able to find in either museum is the type and a small individual, both in the National Museum. The proportions and scutellation of these specimens is given in the following table: = a Thamnophis phenax (Corr), Proc. Acad. Nat. Sci. Phila., 1868, p. 134. a Ott LbAprane yp VARIATIONS OF GARTER-SNAKES, sno Measurements and scutellation of phenax. lig ee TP = | 3 be abe | z 8. % ae a | $ 5 q A Z2|3 ia | a] 2] 8 | se ee Locality. a ‘a = a a < £ a | £8 | Remarks. P a a 4 5 Fa x 2 on > Fa K Pellets | Wee Saw he ie el | S) iS) a = 3) 3 5 ° a ES Z a oa} = e) > nD a a ae mm | mm 30499 | Orizaba, Cordova, | 19-17 8 | 9} 1-3} 161 65 | 605} 130 | 0.22 | Type. Veracruz. | | reg 30498 |....- COwsrmcartrenthanc | 19-17 “| 10 1-3 | 158 73) 275 67 | .24 | Immature. There are no lateral stripes. In the type specimen No. 30499 (which has been in alcohol for many years) the ground color above is brownish olive (said to have been reddish olive in life) with a series of slightly darker cross-bands which extend entirely across the back to the second row of scales. These cross-bands (which are said to have been bright brownish red when the specimen was fresh) are about four scales wide and one and one-half scales apart, and are margined on the longer sides by black borders one-half scale wide, which, owing to the inconspicuousness of the spots themselves, give the snake the appearance of being cross-banded by paired narrow black bars. The nuchal spots are quite similar to the dorsal band in being but little darker than the ground color and heavily margined with black, but they differ from those of the dorsal series in being interrupted on the median line, the posterior black border on either side bending forward to the parietal plates. There is a trace of a dorsal stripe on the nape, between the nuchal blotches. No definite series of spots on the first row of scales. Head brown, marked with black. Supralabials brownish yellow (said to have been light olive in the fresh specimen), and, with the exception of the sixth, partly margined with black. In the second specimen, No. 30498, the pattern differs. from that of the type in that the cross-bars are often interrupted along the median line, in the presence of the spots on the first row of scales, that ante- riorly form a definite row which alternates with the second, and the fact that the nuchal spots are fused across the nape, being strongly notched behind and separated in front by a light line that is continued along the parietal suture to the frontal. Range.—Both of the specimens described above are labeled “Orizaba, Cordova.” The type was listed by Cope as from “Cor- dova, Vera Cruz,” and the locality of the six specimens recorded in 1898 was given as Orizaba, Vera Cruz. The label “Orizaba, Cor- dova”’ seems to mean, not that the specimens came from the towns Orizaba and Cordova, but from Mount Orizaba, in the canton Cor- dova. The range is thus apparently within the range of scalaris, ‘ 132 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. Variation.—lIf the specimens listed in the table are typical of this form, it is interesting to note that the scutellation is somewhat reduced, as shown by the dorsal scale formula of 19-17 and the ten- dency toward the labial formula 7/9. The other scale formulas can not of course be determined on the basis of two specimens, but it is evident that the tail is not long, as in the Sauritus group. Affinities.—As illustrated by the described specimens, the colora- tion of phenaw is strikingly characteristic, so that a much larger series of specimens must be examined before a serious attempt can be made to discover its affinities. Unfortunately we do not have the lateral stripe to aid us, as this is obsolete in the specimens examined, so that we must depend on the other characters. That it is not to be referred to the Radix or Sauritus groups seems to be sufficiently indicated by the fact that the forms of these groups, which occur in Mexico, show no tendency to acquire the phenax type of coloration, although two forms of these groups (proximus and megalops) come very close to or possibly overlap the range of phenax. Furthermore, on the same erounds we believe it to be very improbable that the form in question is to be referred to the Sirtalis group.* The most probable near relative of phenax is, in my estimation, scalaris. Thus, as far as is revealed by the material examined, the scutellation and tail length is exactly similar to that of some speci- mens of scalaris, while the coloration of the former resembles that of the latter, as it does no other in the genus. Thus, in scalaris, as already pointed out, the large transverse blotches encroach upon the stripes which are narrow, and it only requires that these blotches of the opposite sides be continued across the back and the stripes entirely obliterated to produce a type exactly like phenax. It is interesting to note that in one specimen of phenax examined the crossbars are interrupted to some extent along the median dorsal line, producing two series of lateral blotches, as in scalaris, and that the general type of coloration exhibited by these two forms occurs nowhere else in the genus. I believe, therefore, that the two are very closely related, and since they are apparently found in the same localities it is most probable that the so-called phenax speci- mens simply represent individuals in which the process leading to the type of coloration exhibited by scalaris has been continued far- ther to the total obliteration of the stripes and the fusion of the lateral crossbars across the median region of the back. The dis- tinctness of the form should be adhered to, however, until settled by more extensive investigation. a Bocourt (Miss. Sci. Mex., Rept., 1893, p. 778) makes phenax a subspecies of cyrtopsis (eques), thus linking it with the Sirtalis group. I can find no justification for this except the similarity in scutellation, an unreliable factor upon which to base such conclusions, unless it be controlled by others. It will be shown later that eques is represented in this region by one dwarfed form, and that the latter is quite distinct from phenax. VARIATIONS OF GARTER-SNAKES. 133 HA MMONDI.4 Description.—This form, which was long considered a subspecies of elegans, has in late years been given the rank of a species. Any general description will not describe it accurately owing to the variableness of the most distinctive characters. A grayish or brownish yellow stripe on the second or second and third rows. Dorsal stripe usually wanting, occasionally represented on the nape, rarely present for the entire length and then but faintly. Dorsal scale formula, 21-19-17 (occa- sionally 19-21-19-17). Supralabials, 8 (rarely 9). Infralabials, 10 (occasionally 9 or 11). Oculars, 2-3, 2-4, 1-2, or 1-3. Subcaudals, 65-96. Ventrals, 155-178. Proportionate tail length, .21-.27. Color above the lateral stripes dark grayish or olive brown, marked with four rows of small alternating black spots. First row usually a little lighter than above and with or without a small black spot on the base of each scute. Small nuchal blotches. Belly greenish or yellow- ish ash without a median ventral shading formed by the median pro- longation of the black bars at the base of the ventral scutes. Habits and habitat relations (fig. 55).—The only observations known to me on the habits of this snake are those recorded by Van Denburgh and Grinnell: Like other members of its genus, this snake swims well and is usually found in or near water. Its food consists mainly of aquatic animals, such as fish, frogs, and tad- poles. One specimen was caught with a good-sized trout in its teeth. (Van Denburgh 1897, 214.) This is the water snake of the mountain regions, and is abundant in summer along the San Gabriel, Arroyo Seco, and Tujunga canyons. It occurs also tho more spar- ingly in the smaller canyons and out along the water courses a few miles into the valley country. Many years ago before the pumps had drained the water from the Arroyo bed west of Pasadena there was a good-sized stream there all summer. Along this stream the California garter-snake was very common. There were sometimes four in sight at once along the sandy banks. When alarmed they would take to the water and dis- appear into the deeper places for a short time or swim gracefully across the brook and crawl out on the opposite bank. This snake feeds on tadpoles, small frogs, and fish. We have seen a garter-snake so gorged with tadpoles that when alarmed it had to give up some of its cargo, the released tadpoles wriggling out of the snake’s mouth apparently none the worse for wear. (Grinnell, 1907, 49-50.) Range.—This species is only known from southern California and northern Lower California. The specimens examined represent the following localities: La Guilla, San Pedro Martir Mountains, and Sat Antonio, Lower California; Twin Falls, Soda Springs, Los Angeles San Bernardino County, San Diego County, Fort Tejon, Fresno, Alvord, Cartago, Bishop, South Fork Kern River (25 miles above Kernville), Kern River Lakes, Kern Lake, California; ‘‘ Mohave Desert, Arizona.” 9 aThamnophis hammondi (Kennicotr), Proc. Acad. Nat. Sci. Phila., 1860, p. 332. Includes E£. couchi and E. elegans couchi of various authors, but not of KmnNicorr. 33053— Bull. 61—08——10 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. 4 i a ion c & = | = oO a <3) rl [<2] S a =< n ° 4 & [e} a n ) o> = ™ ™ ™ Sie Sin — —_— = eS & rom om = = LQ (6) Ct‘ SE Sr. —_ diagram further shows that there is a considerable range of variation in this character, the extremes approximating those for the genus, e 152 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. for the formula 15-17-15, occasionally exhibited, is the lowest, while the highest, 21-19-17, is next to the highest in the genus. The averages, however, reveal a most pronounced tendency toward a formula less than 19-21-19-17. The average formula for Cali- fornia specimens south of Eureka, California, is about 19-17, while north of this point in California, Greene: and Wechington the average for different localities falls boencer 17-19—17—(15) and 17-15, although in the Puget Sound series the mean approaches 19-17. The variability of the dorsal scale eels is paralleled in the labials. Thus, the supralabials may be 6, 7, or 8, the infralabials 7, 8, 9, or 10. When the averages are taken, Homerer: (figs 64-65), ane onan is seen to be reduced. This is less noticeable in the case of California specimens from localities south of Eureka, in which the average is nearer 8/10 than a lower formula, but north of Eureka, as shown by the table, the frequent occurrence of 6 supralabials and 7 or 8 infralabials draws the mean number of supralabials down to about 7, and that of the infralabials between 8 and 9, the number in both series being slightly higher in the series from Puget Sound. The ventral and subcaudal scutes are also variable, so much so, indeed, that in view of the limited number of specimens examined we can only expect approximate results in the averages. When the larger series are plotted (figs. 66-67), however, the results are strikingly similar to those of the dorsal and labial scutes. Thus, the mean number of ventrals for localities south of Eureka is between 156 and 162, the subcaudals between 76 and 83, while north of Eureka the average number of ventrals lies between 143 and 152, subcaudals between 61 and 70. Again, in the case of these scutes the number in Washington specimens is highest in the Puget Sound specimens. That the variation in the number of subcaudals is not due to a variation in proportionate tail length is shown by the fact that the latter exhibits no apparent differences throughout the range, and is evidently very nearly, if not exactly, the same as in elegans. In the present material the length of the tail varies from .215 (female) to .28 (male), the averages for different localities being between .23 and .26. Even from the few specimens that it has been possible to examine, the above tables show that ordinoides as a form is strikingly dwarfed and that the specimens from the localities north of Eureka and toward the coast tend to have fewer dorsal scale rows, labial plates, ventral and subcaudal scutes than those south of Eureka and toward the Cascade range in Washington. From the present data (fig. 68) ordinoides may have either one or two (rarely three) preoculars. From the table it may be seen that the usual number is one and that the occurrence of two preoculars is only occasional. However, the “SACIONIGUO SIHAONWVHY, NI STIVIAVIVUANI JO YAAWON AHL NI NOILVIUVA AHL DNIMOHS WVYOVIG—'S9 ‘DIY 153 “*purst “USB [BO IaAnoo “YSBM ‘BuLOoB ‘Oos1OURI -UBA‘BIIO} ‘SUIR] ‘WOOdRTIOIS “S010 ‘opis “S0IO uRg ‘YooIg -d1A ‘OYRT -uNoy_ 410q ‘9119BEg Pai @) -vag ‘Arg “BQ ‘qovog 0) [BO me) o1ged uRg *[BO ‘AyuUN0D "TBO xowuog aidwAjQ ‘punog yosng ‘puryiog sayeajvoyg ‘enbdurg Ploy ‘AIO qUedDSAID ‘BYOING ‘OUlDOpUey_ ‘OISBOIN Zig ByuRBS §‘OUuSsal “SACIONIGUO SIHAONWVHL NI SIVIAVIVEdNS 40 HAAWON AHL NI NOILVIAVA FHL SNIMOHS WVUSVIG—'F9 ‘DIA “puvysyt “SBM T8O IaAN0d “USBM “eBuLOdR ‘ooslouRd -UBA ‘BIIOZ ‘SUIBY ‘U00dRTI0I9 : *BaIQ ‘apis 180 UBg ‘Yo. IA ‘oyey -un0y Woy ‘aqyveg “BuO -vag ‘ARg “S010 “3010 ‘AIO 70) 8) ojaeg ueg =‘yeg ‘Aquno0g $89 xouiog oaldurA]Q ‘punog yosng ‘puryiog seyemyeoyg ‘enbduq ‘yoveg ploy yusedseID ‘eyoIng ‘OulOpusy{ ‘OISBOIN Zig ByuUeg ‘OUSeI VARIATIONS OF GARTER-SNAKES. ni tan 154 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. 172 164 156 148 140 Ho Fresno, Santa Cruz Nicasio, Crescent Gold Portland, Puget Comox Cal. County, San Pablo City, Beach, Oreg. Sound, Lake, Cal. Creek, San Cal. Cal. Seattle, Vic- Francisco, Fort toria, Cal. Steilacoon, Van- Tacoma, couver Wash. Island. Fia. 66.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF VENTRAL SCUTES IN THAMNOPHIS ORDINOIDES. 95 85 TI 69 61 55 Fresno, Santa Cruz Nicasio, Cresent Gold Portland, Puget Comox Cal. County, San Pablo City, Beach, Oreg. Sound, Lake, Cal. Creek, San Cal. Cal. Seattle, Vic- Francisco, Fort toria, Cal. Steilacoon, Van- Tacoma, couver Wash. Island. F1G. 67.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF SUBCAUDAL SCUTES IN THAMNOPHIS ORDINOIDES. VARIATIONS OF GARTER-SNAKES. 155 proportionate number of western Washington specimens with two is greater than in the series from elsewhere in the range, while none ve = a of the California specimens exhibit = 5 more than one except those from ° Fresno which have two. It should ecg be stated, however, that two out of Be seventeen “intermediate” specimens é 3 from “Oakland” alsohavetwo. Just ~ sony what this means is difficult to deter- 2 -§ 3 a mine, but it is interesting to note * 225 that the latitudinal variation in the & fa Z = frequency with which the trait—two aa preoculars—occurs in ordinoides is * 8. exactly the same as inelegans. For 5 : some time writers have been led by 2 sty the occurrence of this trait to adopt Pur one of two courses: (1) To separate = E the forms with two preoculars and z Ss class them with similar specimens of 2 pg elegans as a distinct form (biscutata); 2 A (2) to credit the form with having g both a single or a double preocular. [ oe The first view isuntenable, since the nee specimens with two preoculars are = 3 identical with those with one, from 8 sel the same region, while, as I have z ae shown, the western specimens differ 5 markedly in scutellation from the = 2 bm. eastern examples. As I have in- 7 eS S dicated under the discussion of a 2 g 3 elegans, reliance can not be placed = aie on this character as a diagnostic 2 i one, and I believe that the evidence = 5 of the other characters shows con- § 2,42 Gs clusively that in ordinoides, as along 2 23 224 the western border of therange of ele- 2 = 3 2p gans and in the intermediate region * me) between the two forms, the occur- ¥ rence of two preoculars is to be con- Soa sidered as an occasional variation. » ag The color is variable, and so much EB so that it is difficult to determine £22259 to what extent the differences are 2 a : geographic. Nevertheless I believe that the material at hand indicates a generally paler color in the 156 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. specimens from the most western (coast) localities. The most usual color is brown with red markings on the scales; stripes greenish, yellowish or red, the lateral on the second and third rows, the dorsal on more or less of the median three.* On the coast in California, however, specimens are found that have a light olive ground color, no lateral spots visible on the scales, lateral stripes wanting, and a. wide, yellow, dorsal stripe. This lighter color phase thus accompanies the reduced scutellation, a fact which has several times led to the description of such specimens as separate species or subspecies (atrata Kennicott and vidua Cope). They are, it is true, rather unique with their ight color and reduced scutellation, so that, ignoring the variations in these characters, they might easily be considered distinct. Brown’s hypothesis that such individuals represent muta- tions is perhaps excusable on this ground, but with the series before me [ can see absolutely no reason for considering them distinct. In the first place, all gradations occur in the same locality between so-called atrata specimens, with no lateral stripes and a uniform ground color of pale olive, and darker specimens with distinct lateral stripes, while inthe second placea tendency toward the same coloration is evident along the western coast from Santa Cruz County, Cali- fornia, to the Olympic Mountains in Washington. Van Denburgh (1897, 209) states that such specimens occur ‘‘nowhere else but on the coast slope of the San Francisco peninsula.’ That this is a mistake is shown by an apparently typical example from as far north as Crescent City, California, in the Field Museum. This specimen is a typical “atrata” except that the dorsal stripe is red and rather narrower than usual. It is true that no typical ‘“atrata”’ specimens have been found in Oregon and Washington, but this may be attributed to the fact that the specimens from this general region are, as a rule, very dark, nevertheless the two Mount Olympic specimens in the Field Museum, described by Meek (1899, 235-236) as leptocephalus olympia and rubristriata, approach very closely to this type of coloring in being uniformly olive and having the stripes indistinct or wanting. As a rule, however, the ground color of the Oregon and Washington specimens are darker than those from California. At first sight this fact, together with the reduced scutellation of the former, might seem to indicate that we are here dealing with two forms, but as the differences between the two sections are slight, of degree only, and far from constant, I can not consider them sufficient grounds for dividing the form as I understand it. Affimties.—Owing to the position of the lateral stripe upon the second and third rows, ordinoides can be classed only with the Sirtalis aThis is the 7. infernalis of Baird and Girard, Cope, and later writers. The red may also occur in the specimens which are intermediate between elegans and ordt- noides. VARIATIONS OF GARTER-SNAKES. 157 or Elegans groups. I believe that I have proven conclusively in the discussion of the two forms that ordinoides and elegans are not only closely related but that they actually intergrade. This relationship has been more or less recognized as far as the California (elegans) section of the former is concerned for a considerable number of years, but the Oregon and Washington seetion has often been con- sidered as a distinct species. To class this part of the form with sir- talis, as has been done several times, for the reason that the two forms agree in having 19 scale rows and 7 supralabials, is absurd. In the first place, the recognized members of the Sirtalis group wm this part of North America never have more nor less than 19-17 scale rows, 7 supralabials and one preocular, while ordinoides exhibits both more and less than this number of scale rows, and frequently more than one preocular, so that, while the latter is evidently a dwarfed offshoot of some form, the parent stock must have been one with more than 19-17 scale rows and 7 supralabials. In the second place, an undoubted | form of the Sirtalis group occupies almost exactly the same region and probably also the same habitat as ordinoides, which opposes the evidence of a closer relationship between ordinoides and the Sirtalis group. Its affinity with elegans, on the other hand, is shown by the fact that the scutellation and color pattern of ordinoides becomes practi- cally identical with elegans as the range of the latter is approached, the similarity in the scutellation being brought about by an increase in the number of scales in the dorsal, labial, ventral, and subcaudal series, and by the occasional presence of two preoculars. The resem- blance expressed by the similarity in these traits becomes so close in the intermediate region that specimens can not be referred exactly to either form. We believe the conclusion is unavoidable, therefore, that ordinoides represents the geographic extreme of the Elegans group in North America. If this is true it is interesting to note that it is by far the most dwarfed form among the North American mem- bers of the group. CONCLUSION, If it be granted from the evidence produced above that angusti- rostris, melanogaster, scalaris, phenax, hammondi, elegans, and ordi- noides form a group of genetically related forms (fig. 69), this group is very similar to the Radix and Sauritus groups above described in that (1) the form with the largest scutellation occurs on the northern , part of the Mexican plateau; (2) both to the north and south of this center of maximum scutellation a decrease in the number of scales in each series takes place; (3) in North America the group enters a region that is without effective physical barriers to its northward extension and extends as far to the northward as the increasing 158 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. coldness of the climate will permit; (4) the smallest size and scu- tellation is found in the form that constitutes the distal end of the line of genetically related forms; (5) the intermediate region between the forms is narrow in every case, showing that the change in scutellation takes place rapidly. Again, if the relationships are as I have pointed out, there is a remarkable case of convergent evolution in this group. This is shown by the great similarity in scutellation between scalaris and ordinoides. If it were not for the peculiar color pattern of the former, it would be impossible without a knowledge of the locality to distinguish specimens of these forms. ordinoides hammondi \ See \ Te SS angustirostris melanogaster ; NN u s \ \__ | sealaris phenax Fic. 69.—PHYLOGENETIC DEVELOPMENT OF THE ELEGANS GROUP. THE SIRTALIS GROUP (EQUES, SUMICHRASTI, PARIETALIS, SIRTALIS). EQ UES.4 Description.—As in the case of all of the other groups, the maximum ‘ scutellation for the Sirtalis group (as I have defined it) is exhibited by a species which has its principal distribution on the Mexican plateau. This form is the Coluber eques of Reuss, the Eutaenia cyrtopsis of Ken- @ Thamnophis eques (Reuss), Mus. Senckenb., I, 1834, pp. 152-155. Includes E. cyrtopsis Kennicorr, Thamnophis cyrtopsis cyclides Corr, Tropidonotus collaris JAN, E. cyrtopsis ocellata Corr, E. eques aurata Corn, FE. pulchrilatus Corr, E. dorsalis Batrp and Grrarp, and F. ornata Barrp and GIRARD. VARIATIONS OF GARTER-SNAKES. 159 nicott and later writers, Tropidonotus collaris of Jan, etc. It is a very distinct form and has seldom been confused with others. In every specimen examined the lateral stripe is on the second and third rows of scales, and the dorsal stripe is, as a rule, confined to the median row for the greater part of the length, although it is usually wider on the nape and occasionally occupies more or less of the adjacent rows on the body. The dorsal scale formula is nearly always 19-17, rarely 19-17-15 or 17-19-17. The supralabials are 7 (often), 8 (usually), 9 (frequently), or 10 (rarely); the infralabials are usually 10, occa- sionally 11, rarely 8 or 9. The oculars may be 1-2, 1-3, 1-4. Ven- tral scutes 144 (female) to 180 (male). Subcaudals 66 (females) to 97 (males). Proportidnate tail length .204 (female) to .292 (male). The ground color between the stripes is light brown, inclining either to olive or red, and the keels are usually appreciably redder than the rest of the scale. The lateral stripes vary in color from creamy white to light yellow; the dorsal is generally light orange yellow. The first row of scales is ordinarily light or dark ash, with occa- sional, paired dark marks at intervals on the upper and lower edges of the scales, the lower mark including the end of the ventrals. In some instances these marks are represented by a single series of spots, each one of which involves the end of a ventral scute and the first and more or less of the second rows of scales, thus encroaching on the lateral stripe and giving it a wavy appearance. There are always two distinct rows of quadrate black spots between the stripes, and these are generally about one and one-half scales wide, and the same distance apart, the interspaces being bluish white. (According to the descriptions of ‘‘dorsalis” and “‘ ornata” specimens the interspaces in Texas and New Mexico specimens may be red.) These spots are on the skin, and are usually represented only on the extreme edges of the involved scales, except anteriorly, where they are represented by a pair of large conspicuous nuchal blotches which cover all of the included scales with the frequent exception of the keels, which often retain their reddish color. The edges of the scales in the interspaces are usually white. ‘i While the arrangement of the spots described above applies to most of the individuals of egues that have been examined, it does not include a number of specimens in which the spots show a decided tendency to encroach upon the involved scales forming spots similar to those on the skin (ocellata Cope and collaris Jan). In such speci- mens (and these are usually those in which the first row of spots, below the lateral stripe, is better developed) the encroachment of the black pigment upon the scales is generally more pronounced anteriorly. Indeed, it is not uncommon to find individuals with the spots well marked on the scales of the anterior part of the body, but entirely absent posteriorly. Even when distinct on the scales, however, the 160 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. black pigment does not always encroach upon the keels, which fre- quently retain their characteristic rufous tint; but when well devel- oped they often encroach upon the dorsal and lateral stripes, giving them a wavy or interrupted appearance. In a number of specimens which have been examined (mostly from the northern part of the range, see below) the spots in the second and third rows fuse on the anterior part of the body to form distinct cross-bands between the stripes, occasionally also uniting with those of the first row, thus intersecting the lateral stripe. This tendency has not been noted to take place for more than seven spots posterior to the nuchals, which in these specimens are usually fused across the nape and strongly notched posteriorly by the dorsal stripe. Belly grayish white, light yellow or light greenish yellow. The usual narrow, black bar, and ventral spots may occur alone or together, but in any case rarely show much beyond the edges of the superincum- bent scute. The ventral plates are rarely speckled with darker. The head above is usually dark brownish olive splashed with black. The supralabials (with the exception of the last), lower and middle post- oculars, and the lower part of the preoculars are usually pale yellow; supralabials well margined with black. The last labial is usually more dusky, approaching the body color. Generally many of the infralabials are also margined with black. Halits and habitat relations.—Apparently the only observations that have been recorded upon the habitat preference of this snake are those on the three specimens taken in Sabino canyon, Santa Catalina Mountains, Arizona, in 1907 (Ruthven, 1907, 588). ‘‘All of the speci- mens obtained were found near the stream in the canyon. Two were lying on rocks in midstream, the other on the immediate shore. When frightened they immediately took to the water, swimming in the swift current with apparent ease. They doubtless subsist largely upon the frogs and tadpoles which abound in this habitat.” Bailey records it in the Davis Mountains, Texas, at an altitude of 5,700 feet. Range.—T. eques is known to range over more degrees of latitude than any other garter-snake that inhabits Mexico, except proximus. The range extends from central Guatemala on the south to the high plateau in central Arizona and New Mexico on the north. Specimens have been examined from the following localities: ‘‘Central Guate- mala,” Escuitla, Guatemala; valley of Mexico and Toluca, Mexico; Veracruz, Zacaultipan, Hidalgo; Guanajuato; Rinconado, Puebla; Durango and Coyotes, Durango; Fort Apache, Fort Huachuca, White River Canyon, Sabino Canyon (Santa Catalina Mountains), Fort Whipple, Arizona; San Ildefonso, Lake Valley, New Mexico; Davis Mountains, Pecos, Helotes, San Antonio, Texas. The literature on this form is rather extensive, but, as far as I can find, no specimens have been recorded outside of the region outlined — VARIATIONS OF GARTER-SNAKES. 161 by the above localities, except the specimens listed by Boulenger (1893, 209) from Jalisco and Guerrero and the specimen said to have been taken by Xantus at Cape San Lucas, Lower California. 1 have elsewhere shown (Ruthven, 1907, 588-589) that there is undoubt- edly a mistake in the latter record, in that the specimen upon which os NE ee 5 we SK Se Ne . ¥ 4 s 4¥ z i aA, be be |. : roan Sx x RO BGie iss aS Pa Pipitege me ‘ eR ato tt 6 Fic. 70.—DISTRIBUTION OF THAMNOPHIS EQUES, AS INDICATED BY THE LOCALITY RECORDS. it was based very probably came from Arizona. As known at pres- ent, then, the range of 7. eques may be defined in general as all of Guatemala, Mexico exclusive of the coastal plains, the Proplateau region of southern Arizona and New Mexico, and the high plateau region in the last two named States and western Texas (fig. 70). The 162 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. specimens in the U.S. National Museum labeled ‘‘ Helotes”’ and ‘‘near San Antonio,’ Texas, seem to indicate that it also occurs in the prairie region of central Texas, but as these localities are near the margin of the high plateau the range of the species can not be ex- tended into the prairie region until notes on its occurrence in this region are at hand. Just how far north it extends is questionable, but the San Ildefonso specimens show that it attains the high plateau region in New Mexico, which might be expected, since it occurs in the mountains in the Proplateau region (Bailey, 1905, p. 48; Ruth- ven, 1907, 588-589). Similarly its northern limit in Texas also remains to be determined. Variation.—Naturally in such an invariable form it is difficult to discover geographic differences in any but large series of specimens. Some differences appear in the material examined, however, which seem to be associated with definite parts of the range. In the dia- grams (figs. 71 and 72) I have plotted the variations and mean number of dorsal scale rows and supralabials for various localities throughout the range. It is readily granted that the number of specimens employed in the tables is inadequate to furnish conclusive results in an investigation of the geographic variation in this form, and yet,in view of the nar- row limits of variation, I believe that the tables are not deceptive in indicating a slight decrease in the number of dorsal scale rows and supralabial scutes in southern Mexico and Central America. The individual variation is too great in the case of the infralabials, ventral and subcaudal scutes to insure accurate results when the available data is tabulated, but I believe when more material is available that the evidence of these characters will not vitiate the evidence of the dosal scale rows and supralabial plates that there is a reduction in scutellation in the region mentioned. The tendency toward a reduction in the number of dorsal scale rows in the southern part of the range is shown in the occurrence of the formula 19-17-15 and 17—19-17—15 in specimens from this region, which brings the aver- age below 19-17; while similarly the decrease in the number of supra- labials in the same general region is shown in the frequent presence of 7, a number which is not shown in any specimens north of the State of Durango. The latter fact justifies me at once in relegating Cope’s spe- cies pulchrilatus (1885 b 174) to the synonomy of eques, as this form was based entirely upon specimens of eques with 7 supralabials. Cope (1900, 1062) objected to this disposition of the form by Boulenger with the statement that it ‘“‘belongs to a different section of the genus.” Tle neglects to mention, however, which ‘‘section”’ it is to be referred to, and there is no other evidence besides the number of supralabials that I can find to separate it from eques. The limits of variation in the ventral and subcaudal scutes and tail length have already been given. The averages for the larger a3 —— © VARIATIONS OF GARTER-SNAKES. 163 series in which males and females are represented in about equal pro- portions are about as follows: Ventrals, 160-170; subcaudals, 75-85; tail length, .23; but much larger series are necessary to properly define these characters. _ As in the case of scutellation, there is little marked variation in the coloration. However, the specimens in which the dorsal spots 419417 i is a if es ze feat b> ea lel Guatemala, Valley of Guana- Durango, Fort Hu- SanIlde- Davis Helotes, Cent. Amer. Mexico, juato, Mexico. achuea, fonso, Mts., Texas. Mex. Zacaulti- Arizona. NewMex- Texas. pan, ico. Mexico. Fig. 71.—DIAGRAM SHOWING THE VARIATION IN THE DORSAL SCALE FORMULA IN 'THAMNOPHIS EQUES. tend to become distinct on the scales and fused into cross-bars ante- riorly (see p. 159) are principally from the southern part of the United States. Cope (1880, 22-23) has described these as a distinct variety ocellata (and Jan has made them the basis of the species collaris). This pattern is found in occasional specimens from nearly every Guatemala, Valley of Guana- Durango, Fort Hu- San Ilde- Davis Mts., Helotes, Cent. Amer. Mexico, juato, Mexico. achuea, fonso, Texas. Texas. Mex. Zacaulti- Arizona. New Mex- pan, ico. Mexico. Fia. 72.—DIAXGRAM SHOWING THE VARIATION IN THE NUMBER OF SUPRALABIALS IN THAMNOPHIS EQUES. locality in the range of eques, and while of more frequent occurrence in the northern part still occurs only as an individual variation of the normal coloration. In regard to the loss of the stripes, the case is somewhat different. Thus, in the series in the U.S. National Museum from Zacaultipan, Hidalgo, Mexico, out of six specimens three are typical, eqgues, and the others, while agreeing with them in scutella- tion, differ in the apparent total absence of lateral and dorsal stripes. 164 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. The latter three specimens have been referred by Cope to sumi- chrasti, and conform closely to the original description of that form. Inasmuch, however, as it is impossible to work out the dorsal scale formula of the type (owing to its poor condition) the identity of the Zacaultipan specimens with them can not now be determined. The species was described, however, from specimens from Orizaba, Mexico, and credited with 19 scale rows, which probably means the maximum number, and since specimens from that locality have generally a more reduced number of dorsal scale rows (although the maximum is frequently 19) and a similar coloration, the types of swmichrasti are undoubtedly identical with such specimens which are to be referred to a different form. I believe the specimens from Zacaulti- pan, without stripes, can be considered only as variations of eques, which possibly indicate that the form tends to lose the stripes in this region.@ Affinities —If I have proven that there is a tendency toward a reduction in the dorsal scale formula below 19-17, and toward the obscuring of the stripes, the only objection to the conclusion that eques is very closely related to the following form (swmichrastr) has been eliminated. As will be shown in the description of the latter, there are no appreciable differences between so-called swmichrastr specimens from Orizaba, Veracruz, and typical eques except in the slightly larger number of scale rows of the latter; a discrepancy that entirely disappears when it is noticed that in the intermediate region, geographically, the scutellation of eques tends to become apparently exactly that of swmichrasti. On the north we believe that it is directly related to parietalis (see p. 172). SUMICHRAS TI. Description.—l use this name to designate those specimens, so far recorded only from southern Mexico, that otherwise combine the usual coloration of eques with a reduced scutellation and the usual absence of the dorsal stripe. 21 can find absolutely no character in which the types and only specimens of Eutaenia aurata Corr (1892, 659-660) and Thamnophis cyrtopsis cyclides CorE (1862, 299) differ from typical eques specimens. b Thamnophis eques swmichrasti (CopE), Proc. Acad. Nat. Sci. Phila., 1866, p. 306, Includes also E. chrysocephala Cope and Tropidonotus godmani GUNTHER. I use this name advisedly. As before stated, the types of swmichrasti are in very bad con- dition, but they are apparently very similar in coloration to so-called chrysocephala specimens from the same region. Cope always distinguished the former as having 19 scale rows and the latter as having 17, but since typical chrysocephala specimens with 17-19-17 scale rows are found inVeracruz, the distinction is not a sufficient one. Since also both forms are characterized by the usual obscurity of the dorsal stripe, I feel justified in concluding that it was upon specimens of this kind that Cope based his description of swmichrasti, which name should, therefore, be the name of the form. Tropidonotus godmani was evidently based on a specimen of this form with a dorsal stripe. VARIATIONS OF GARTER-SNAKES. 165 The writer has seen but eleven specimens of this form (inclusive of the damaged types of swmichrasti), but these are so uniformly unique that he has no hesitancy in giving them the rank of a form The lateral stripes are generally present, although occasionally not very distinct, and are in every specimen on the second and third scale rows anteriorly, but generally only on the second posteriorly. The dorsal stripe is usually absent, although occasionally present for a short distance anteriorly, more rarely for the entire length. Coloration otherwise much as in eques, the ground color being brownish with lighter keels, and the lateral spots little evidenced upon the scales, except anteriorly, where they form a pair of distinct nuchals and occasionally one or two large transverse blotches between the stripes. The scutellation has been summarized in the following table: Scutellation of Thamnophis sumichrasti Cope. | ra = hal — aoe ay a a a af | Bans Foe ieee a | 3 PaaS No. | 25 | s |S| els | se LWASE | Locality. a Sia leith cee ok Sees, a | es Remarks. N.M. 3 tlic Sha foes cel eileen eh H hl dee |] asi Ssh eal dena || oes m' | os © ill =) |) o s S) 3 BP | A SO |e ken: rete inex aleaa een | mm mm 26501 Orizaba, Mexico) CO) RG) tesa Ce I CO GOL sl el Coy || abhor 26502 dome see: (*) 8 | (%))| 13 COME NC Oa ©@ 7077 Owe eee. 17-15 8 | OV eS MEK @Qyll TQM Oy ay MOManl\eeees Ome aes: 17-15 | 8| 10) 1-3} 149} (2) €?) | (2) | (% | TOAD eee Gomer ee oe 17-15 8 | UO | Teas) ER (Cay COL | ay) Le) Vr AVC bee oe say eee 17-15 | s i {is 149 | 76 | 496 | 126 0.254 | | | | 5 | 1. If 9| | | | | ZOvZde aeeee Comet eee 17-15 | 8 tio 1-3 | 147] 74| 425 | 108 | .242 | | | | | MOWenleaee dot sence | 17-15 | 8 | 10 | 1-3 | 152] 83] 223| 55 | .246 MOUE | 5 2128 dO’ sett ee | 19-17-15 | 8} 10] 1-3] 155 |(?)} (%) 1 (2%) | @) | | 9 | | 30494 |... dope eeeee: | 17=15 | lio lis 153 | 80] 530 | 136 | .256 | Type of chry- Ve eras ae | socephala. 30496... dO ore ea. 17-19-17-15 | 8 | 10 te | 154./(?)) (2) | (2) | @) | } zi | As shown from this table, the form sumichrastt as now known is characterized by a small dorsal scale formula, and exhibits an appar- ent tendency toward a smaller number of infralabials than 10, and possibly also a smaller number of ventral plates than is usual in the larger members of the genus. The tail length as far as shown by the material is similar to that of most of the forms exclusive of those of the Sauritus group. Range.—The only specimens of swmichrasti that I have seen are from Orizaba, Veracruz. Boulenger (1893, 203) records specimens of chrysocephalus from Omilteme and Amula, Giterrero, Giinther (1894, 133) specimens of godmani from the same localities, Bocourt a specimen of swmichrasti from Coban, Guatemala, and Cope a specimen of swmichrasti as possibly from ‘‘the Plateau of Costa 33553—Bull. 61—08——12 166 BULLETIN -61, UNITED STATES NATIONAL MUSEUM. Rica at Cartago.”’ Needless to say the dorsal scale formula of these specimens should be carefully examined to make sure that they are are not eques specimens with obscure stripes. As already stated, it seems advisable to refer to eques the specimens from Zacaultipan, Hidalgo, determined as swmichrasti by Cope. Affinities —As already mentioned under the discussion of eques, I believe that these two forms are directly related. They are remark- ably similar except for the smaller dorsal scale formula and usual obscurity of the dorsal stripe in swmichrasti, and since I believe it demonstrated that there is a tendency toward a reduction in the number of dorsal scale rows and an obscuring of the stripes in eques in southern Mexico, there can be no objection to considering the individuals from Orizaba, Mexico, as representing the continuation of this process. The form apparently lies entirely without the province of eques, and intergrades with the latter through such specimens as the ones from Hidalgo, valley of Mexico, etc., referred by Cope to swmichrastt and pulchrilatus. Just where the intergra- dation occurs can not now be determined, but it should be noted that the difference in the number of scale rows must be brought about in quite a narrow area, since the formula 19-17 is quite constant in eques, even in the States of Hidalgo and Mexico, where it comes close to the range of suwmichrastt. PARIETALIS.« Description.—This form is so well defined that it has seldom been confused with any other, and has avoided the fate of so many other forms, 1. e., being broken up into a number of subspecies. The dorsal scale formula is always 19-17; the supralabials usually 7, frequently 8, but very rarely 6; the infralabials usually 10, occasionally 9, more rarely 11, and very rarely 8; the ventral scutes 150 to 178; subcau- dals 65 to 92; tail length .202 to .32. The general coloration is as follows: Ground color above dark olive to brownish olive or dull reddish brown, the color being confined to the scales between, and the keels of the scales involved in, the lateral spots. The latter are arranged in two rows on the skin between the lateral and dorsal stripes; occasionally the spots in both rews distinct, but those of the upper row usually fused for more or less of their width to form a black band, the spots of the lower row appearing as down- ward projections from this band. The black pigment of the spots only encroaches upon the edges of the involved scales. The inter- spaces on the skin vary from orange to red, and this pigment nearly always encroaches on the edges of the involved scales. The first few @ Thamnophis sirtalis parietalis (Say), Long’s Exped. Rocky Mts., I, 1823, p. 186. Includes Eutexnia sirtalis trilineata Cork (part), E. sirtalis tetratenia Cork (part) (see also p. 176), and £. sirtalis dorsalis Cops (not E. dorsalis BARD AND GIRARD; see p. 158). VARIATIONS OF GARTER-SNAKES. 167 spots anteriorly of the two series may fuse to form transverse blotches between the stripes. The dorsal stripe is on the median and halves of the adjacent rows and with the lateral stripes may be bright orange yellow, bright yellow, various shades of greenish yellow, or bluish. The first row is usually sufficiently darker to define the lateral stripe below, but this is not always the case. The belly may be grayish, greenish, or bluish. The better marked variations from the normal type will be considered under ‘‘variation.”’ Habits and habitat relations.—Few definite observations have been recorded on the habits of this form. Stomachs examined frequently show specimens of the leopard frog (ana pipiens), and Taylor (1892, 325) writes that in Nebraska ‘‘specimens of this garter, not exceeding two and one-half feet in length almost always contain within their stomachs specimens of the common earthworm.” Specimens in cap- tivity will eat both frogs and toads voraciously. In northern Iowa parietalis is rather rare. On a two months’ col- lecting trip in Clay and Palo Alto counties, July and August, 1907, an expedition from the University of Michigan Museum obtained but three specimens, although a careful search was made for them. Of these two were found on the immediate edge of large sloughs (fig. 23), the third in a rather swampy swale on the prairie. We have also, however, taken a few specimens in the long grass on the uplands in this region (fig. 24), but they are apparently very rare in this habitat. Another specimen was taken on the bank of a small creek near the Missouri River, in Woodbury County, lowa. One of the specimens taken in Palo Alto County in 1907 was a large female, and on Septem- ber 30 gave birth to seventy-three young. Range.—lIt is rather interesting that, although parietalis ae been well known for many years, the fate of its range can nowhere be established at present except on the west. Its range is greater than that of any other known garter-snake and includes a numper of biotic regions. It may be said in general to extend from the western part of the ‘‘prairie peninsula’? in Iowa to the Cascade range in Oregon, Washington, and British Columbia, and to the Pacific coast in Cali- fornia, and from northern New Mexico into Canada. Undoubted specimens have been examined from the following localities: Menio Park, Fort Reading, Pitt River, San Francisco, Petunia, Fort Crook, Fresno, Crescent City, Tomales Bay, Petaluma, Eureka, Yosemite Valley (altitude 4,000 feet), Palo Alto, Camp Bidwell, California; Gold Beach, Des Chutes River, Dalles, Warners Second Lake, Oregon; Cheney, Pullman, Fort Walla Walla, Washington; Sicamous, Nelson, Kaslo, Donald, British Columbia; Lake Tahoe, Camp 10 and 12 (Ridgway), Nevada; Carson Valley, Logan (10 miles east of), Utah; Fort Collins, Denver, Greeley, Colorado; Flathead Lake, Swan Lake, Fort Benton, Bitter Root Mountains, Billings, Fort Custer, Three 168 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. Forks, Fish Creek, Montana; Peabody, Onaga, Fort Riley, Fort Harker, Kansas; Platte River, Fort Kearney, Nebraska; Ames, Des Moines, Palo Alto County, Clay County, Woodbury County, Iowa. Further collecting will extend the range somewhat beyond the extreme points listed above, to the north, south, and east. At pres- ent the most debatable territory is to the south and east, for here parietalis meets two allied forms (eques and sirtalis), which it resembles so closely that it is impossible to distinguish preserved specimens with certainty. This is true of specimens from northern New Mexico and Arizona, and Minnesota, eastern Iowa, and western Illinois. The points of resemblance between these forms will be discussed under ‘‘variation,’’? but it should be stated here that no reliance can be placed upon the published lists which record parietalis for any of the last named States. Thus, Cope records it from Lake Valley, New Mexico, but the basis of the record proves on examination to be a specimen of eqgues, while apparently several of his dorsalis specimens from the same state, as well as Garman’s Illinois specimens and Ruthven’s Isle Royale specimens (1906, 111-112), are not at all typical and are mostly to be considered ‘‘intermediate.”’ At the present time the eastern limit of typical parietalis may be given very generally as the ‘‘prairie peninsula” in central and southeastern Minnesota, eastern Iowa, and northeastern Missouri. The southern limit in central United States can not even be approximated, although it doubtless lies in the northern part of Texas, Arizona, and New Mexico. The southernmost point from which we have examined specimens from California is Fresno, but Van Denburgh (1897, 203) records it from ‘‘San Bernardino (Ontario) and Riverside (Riverside) counties,’’ and Grinnell (1907, 49) from the Pacific slope of Los Angeles County. The known range has been platted: on the map (ie. 73). Variation.—Like its relatives, eques and sirtalis, parietalis is little variable in scutellation. Large series alone, therefore, will reveal geographic differences if they exist. Unfortunately, while I have examined about three hundred and fifty specimens, these have been from many localities, so that single regions are represented only by small numbers. On the other hand, in view of the constancy of the characters, geographic variations must be very slight, and I believe that the data accumulated is quite representative of the actual con- ditions. In every specimen in which they were counted (311) the dorsal scale formula is 19-17. In regard to the labials, the suites are all alike in that most of the specimens have the labial formula 7/10, but quite frequently 8 supralabials and 9 or 11 infralabials. I have seen but four specimens in which the variation exceeded this amount, one from Eureka, California (Field Museum, 1111), and three from Pullman, Washington (Ruthven collection), each of which VARIATIONS OF GARTER-SNAKES. 169 have 8 infralabials on one side, and one from Pullman, Washington (Ruthven collection, No. 11), which has 6 supralabials on one side.¢ The extent of the variation in the labial formula may be seen in the diagram (figs. 74-75), for the Montana specimens. The conditions @]t is worth recording here that a female from Pullman, Washington, which the writer kept in captivity, and which had a labial formula 7/10, gave birth to eleven young none of which had more than 9, while three had 8 infralabials on one side, and none had more than 7, and one had 6-7 supralabials. There is, in regard to the labials, more reduced specimens in this brood than in the series from any other locality. The difference is not geographic, as six specimens from Walla Walla have 7/10, with one exception, in which there are 9 infralabials on one side. All that can be said is that the cause that brought about the reduction apparently affected the entire brood. .—DISTRIBUTION OF THAMNOPHIS SIRTALIS PARIETALIS, AS INDICATED BY THE LOCALILY RECORDS. ae Fig. 170 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. represented in these diagrams are typical for every locality examined, as far as can be judged from the material at hand. The ventral scutes vary as follows: Females, 150-165; males, 153-178. The average for thirteen specimens from California (west of the Sierra Nevadas) is 161, for twenty Montana specimens, 163.7. The subcaudal plates vary as follows: Females, 65-80; males, 73-92; average number for twelve specimens from California, 82; for thirteen specimens from Montana, 81.3. The material at hand is not sufficient to throw much light upon the proportionate length of tail. The extremes are .202—.262 (female), .239-.32 (males). The average for different localities with more than twelve specimens falls between .23 Percentage of Percentage of specimens. specimens. .6 0 .90 64 48 eae 16 .O0 00 Supralabials. Infralabials. Fi1Gc.74.—V ARIATION IN THE NUMBER OF SUPRALA- Fic. 75.—VARIATION IN THE NUMBER OF IN- BIALS IN 20 SPECIMENS OF THAMNOPHIS SIRTA- FRALABIALS IN 20 SPECIMENS OF THAMNO- LIS PARIETALIS FROM MONTANA. PHIS SIRTALIS PARIETALIS FROM MONTANA. and .25, but to what extent these averages are constant throughout the range is a question which awaits investigation. The more striking variations in color in parietalis are principally in the way of modifications in the arrangement of spots, the intensity of the ground color, and the color of the interspaces on the skin. It should be recalled that the usual coloration is: Ground color, olive or reddish brown; upper row of spots fused on skin; interspaces about one scale wide and red in color; dorsal stripe on the median and halves of adjacent rows. The variations are as follows: (1) Ground color above, black; interspaces, narrow (from three- fourths to one-half the width of a scale); dorsal stripe on the median —--- VARIATIONS OF GARTER-SNAKES. ial and halves of the adjacent rows. Such specimens are characterized only by the slight excess of black pigment. We have seen this color phase only in specimens from western Montana (Fort Benton, Flat- head Lake, Three Forks, Bitter Root Valley) and the west slope of the Rocky Mountains in southern British Columbia. It was princi- pally on parietalis specimens of this type that EF. sirtalis trilineata Cope (1892, 665) was based, but the characters are not sufficient to warrant the separation of these specimens from parietalis. They are indistinguishable from some specimens of concinnus (see p. 176). (2) Quite often in typical parietalis specimens the red pigment of the interspaces shows a tendency to encroach upon the black areas, particularly between the fifth, sixth, and seventh scale rows, to the partial or total separation of the lower row of spots from the upper black band. This is not at all remarkable, since it is these scale rows which mark the limits of the spots in garter-snakes which have the two rows distinct. In two specimens from Pitt River, California (Cat. Nos. 21383 and 21384, U.S.N.M.), the separation of the two rows of spots is complete, and a further modification exists in the fact that the lower row of spots, like the upper, has also fused into a longitudinal black band. These specimens have been described as E. surtalis tetrataenva Cope (Yarrow, 1875, 546), but they are, in my judgment, evidently only two anomalous specimens of parietalis, and this conclusion is supported by the fact that in one of these specimens (No. 21383) the lower black stripe is partially broken up into spots. A similar pattern is occasionally exhibited by concinnus (see p. 176). (3) A number of specimens of parietalis from Kansas and Minnesota examined have the first few spots anteriorly fused into transverse blotches between the stripes, as in some Illinois specimens of sirtalis described by Cope as Hutaenia sirtalis semifasciata, and in the eques specimens described by him under the name ocellata. This is merely an individual variation, but may have some significance (see p. 173). (4) By far the most significant variation exhibited by parietalis is the breaking up of the lateral black band into the usual upper row of spots. If this alone takes place the result is a coloration practically the same as that of eques and sirtalis, except that the interspaces are red. We have seen this color phase in Colorado (Greeley) and eastern Kansas (Onaga) and Isle Royale, Michigan, specimens, but it can only be certainly determined in fresh material, for the red of the interspaces fades rapidly in preserved specimens, so that they can not with cer- tainty be distinguished from eques (when there are 8 supralabials) and surtalis, the more so that the modification seems to occur only near the ranges of these forms. In Minnesota, eastern Iowa, and Missouri the proportion of speci- mens with two distinct rows of spots is large, and the specimens from this region are further characterized by the frequent absence of red pigment on the sides, a character which is also occasionally found in 172 BULLETIN 61, UNITED STATES NATIONAL MUSEUM, Kansas and Nebraska specimens. Just what influences the presence of this rea pigment can not now be determined, but it is interesting to note that in a series of sixty-five young born in captivity of a typical parvetalis mother (upper row of spots fused and red interspaces) from Clay County, Iowa, but one showed the least trace of red at birth, and this one but little, showing that the red color may be developed to a large extent after birth, as in most North American birds. Eastern specimens with two rows of spots and red interspaces have been examined from eastern Kansas, eastern Iowa, eastern Missouri, and Isle Royale, Lake Superior. The Colorado and New Mexico material in which specimens with the same arrangement of spots occur were referred by Cope to dorsalis. Affinities. —One need not search far for the relatives of this form. There can be no question but that to the eastward it grades into sirtalis, and although, as before stated, typical specimens of parietalis may be found as far east as eastern Iowa and sirtalis specimens as far west as Kansas, the section of this form which inhabits the entire prairie region (i. e., east of central Kansas and Nebraska) as well as the edge of the adjoining forest region * may be considered as intermedi- ate, for in this region the distinctive characters which separate typical parietalis (upper row of spots fused and red interspaces) break down. The presence of specimens in Colorado with two distinct rows of spots is alsosignificant, inasmuch assuch individuals are indistinguish- able from eques, when the red of the sides has disappeared, for we have already noted that egues may have 7 supralabials, and apparently shows a tendency to have red interspaces in Texas and New Mexico. This is why it is impossible to define at present the respective south- ern and northern limits of the two forms, points which can not be settled until their relations are worked out from fresh material. Since, however, in the two forms the scutellation (with the exception of the number of supralabials), proportions, and position of the lateral stripes are approximately the same, while the ranges are at least near together, the evidence seems to indicate a close relation- ship, and we believe that the similarity in coloration of specimens from near the common boundary is such as to indicate an actual intergrad- ing. As already noted, the number of infralabials is noncommittal in small series, as they may be 7 or 8 in both forms. Specimens have been examined from as far south as Taos, San Ildefonso, Las Cruces, and Albuquerque, New Mexico, that can be referred to parietalis, as well as others from San Ildefonso that are apparently eques, but this does not signify that intergradation does not occur. The whole question must await further investigation, al hes e eee fea that on Isle Renate Pee whic h is Fai Pinte the forest region, most of the individuals are typical sirtalis in coloration, but that speci- mens are also found in which the lateral spots are in two rows and the interspaces bright red, "' wr. = (ee) oe . VARIATIONS OF GARTER-SNAKES. ive when fresh specimens shall be available from this region. At present we can only say that the forms are undoubtedly closely related, and belong to the same group.” In the northwest parvetalis meets another form (concinnus), with which it is closely allied. The relationship of these forms will next be discussed. CONCINN US. Description.—Doctor Hallowell (1852, 183), in his description of this form, concludes with the remark that it is ‘‘The most beautiful of the North American serpents hitherto discovered.’ His enthusiasm is easily realized by anyone who has the good fortune to see living specimens. The form differs from parietalis only in the increased amount of black pigment and the attendant effect upon the pattern. The dorsal scales, including the first row, are generally black, with the exception of more or less of the scales involved in the interspaces, which are red. The stripes may be brightly colored, but are usually yellowish, greenish, or bluish. The lateral when present is on the second and third scale rows, but is frequently absent; the dorsal is generally confined to the median row, but may also cover the halves of adjacent rows or be entirely obsolete. The spots when distinct are arranged as in parietalis, i. e., those of the upper row fused, those of the first row distinct from each other but fused with the upper row. The interspaces (always red) are, when present, however, generally much narrower (being only one-half to one-fourth a scale wide) than in parietalis and are frequently entirely absent, the skin being solid black. Belly usually dark greenish or bluish, and gener- ally much speckled with black and occasionally orange. It will be seen from the above description that there is no character which will constantly distinguish specimens of concinnus from parie- talis. The narrow dorsal stripe and lateral interspaces of the former will usually do so, but these may be exactly as in parvetalis. Still, the fact that nearly all specimens from Washington and northern Oregon, west of the Cascade range, are characterized by a marked predomi- nance of black pigment and a narrow dorsal stripe justifies their recognition as a separate form. Habits and habitat relations.—As far as I know very little has been recorded on the habits of this snake. Cooper and Suckley (1860, 297) record the following notes: This species exists in great abundance near Steilacoon and Nisqually. They are found on the gravelly prairies and in the vicinity of the numerous small lakes of this section of country. About the first of April they begin to come out of winter @ It is interesting to note that the three forms, eques, parietalis, and sirtalis, all exhibit the occasional tendency toward the fusion of the first few anterior spots into cross-bars. b Thamnophis sirtalis concinnus (HALLowELL), Proc. Acad. Nat. Sci. Phila., 1852, pp. 182-183. Includes Eutenia pickeringi Barrp and Grrarp, L. sirtalis trilineata Cork (part), HZ. sirtalis tetrataenia Cork (part). 174 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. quarters, and can be seen at midday sunning themselves near small clumps of scrub- oak bushes, to which they retreat when alarmed. A little later they are found in couples or in small companies. Although they are rarely ever found more than one- eighth of a mile from water, they are, nevertheless, still more anxious to be close to it as the season advances. They will then (in May and June) be found lying close to the water, on the lake shores in the grass, and among the sedge of the marshes, and even upon.small bog islands, as much as 50 yards from the shore. In summer, like other members of this genus, they are found lying in small pools and in the water at the edges of the lakes during the heat of the day. They appear to be a harmless, lazy species, and, as above stated, exceedingly fond of the water. wi “ee ‘ STAIN wl \f os FIG. 76.—DISTRIBUTION OF THAMNOPHIS SIRTALIS CONCINNUS, AS INDICATED BY THE LOCALITY RECORDS. Range.—The range of concinnus is apparently rather definitely bounded on the east and west, since it ranges from the Pacific Ocean to the Cascade range. Specimens which I have referred to concinnus have been examined from the following localities: Portland, Eugene City, Oregon; Shoalwater Bay, Fort Vancouver, Tacoma, Puget Sound, Seattle, Fort Steilacoon, Olympic Mountains, Port Angelus (30 miles from), Lake Washington, Washington; Comox Lake, Vic- toria, British Columbia (fig. 76). Variation.—Judging from the available material, concinnus is identical with parietalis in scutellation and proportionate tail length. VARIATIONS OF GARTER-SNAKES. 175 The variations and averages in these characters for nearly all of the specimens examined are placed in the following table: Wer S. labials. I. labials. Ventrals. | Subcaudals. | Tail length. or- 35 c ; : Locality. ee —————e | ai : Sais. Range. Ave. | Range. Ave. Range. | Ave. Range. Ave. | Range. | Ave. | | | | | | (24 specimens.) Seattle, Taco- | | ma, Puget Sound, Wash- | AOS Meee 19-17) 7-8 | Fort Steilacoon, | | | | =I _ to on nS) 8-10 9.8 152-170 163 | 63-90 | 76.5 |0. 215-0. 286 | 0. (3 specimens.) Port Angelus, | | | Ody. ni pric . Mountains, Washington. ..} 19-17 TS) || Yer 9-10 | 9.7) 155-165 | 162 79'| 79 -20- .259 | .238 (7 specimens.) Comox, Victo- | | ria, Vancou- | | | | ver Island...-.- 19-17 7-8 i. || 8-10 9.4 160-170 164 69-86 76 .219- . 266 . 238 (9 specimens.) | Portland, Ore- POM. 2eeeee 19-17 7-8 .2 | .208- .265-| .242 “I — 68-86 ~I “I 910} 9.8] 152-172) 164 The variation in color has been briefly treated in the general descrip- tion. There are two color phases, however, that are quite well marked. (1) Stripes usually present, the dorsal generally confined to the median row. Spots of first row either distinct or not differentiated. Belly blackish. This is the type of color described by Baird and Girard and since known as pickeringi. It is found in western Wash- ington and to the northward. (2) Ground color black. Lateral stripe absent. Interspaces (red) present. Dorsal stripe occupies the median and adjacent half rows. Belly blackish. This is Tropidonotus concinnus Hallowell. So far - the only specimens that we have seen hail from Portland and Eugene City, Oregon, and it is interesting to note that out of nine specimens from this immediate region (in the U. S. National Museum) every one is of this type. Notwithstanding the fact that it has thus apparently a different range from color phase (1), I am in harmony neither with those herpetologists who consider this color phase as an individual variation of parietalis nor with those who would give it subspecific rank. To meit seems that its distinctive traits (width of dorsal stripe, together with a loss of the lateral) are too slight for the latter posi- tion, and that the increased amount of black pigment classes it dis- tinctly with the melanistic form that inhabits western Washington 176 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. and the adjoining part of British Columbia; in which case. the name concinnus Hallowell (1852) takes precedence over pickering: Baird and Girard (1853). As stated under parietalis, variations in color occur which are indistinguishable from some of those of that form. Thus, the black pigment on the sides may be disposed as two black bands separated by a red one, and it was upon a specimen of this kind from Puget Sound, Washington, as well as the two from Pitt River, California (p. 171), that Cope based his subspecies tetrataenia. Similarly, when the dorsal stripe occupies one and two half rows, as it occasionally does, the specimens are very similar to dark specimens of parietalis from Montana, and in fact one of the type specimens of trilineata was a concinnus from Fort Townsend, Washington. Affinities —Concinnus grades directly into parietalis through dark specimens of the latter (the phase trilineata) in the Cascade range. It has no direct relationship with ordinoides, as may be seen by the scutellation, for we have shown that ordinoides, has generally less than 19-17 scale rows, while the variations above this number in the latter indicate that it is derived from a form with a larger formula than 19-17. ; SIRTALIS. a Description.—This form is one of the best known in the genus, probably owing largely to the fact that it is very common in a thickly settled region. At the present time the material available for study numbers nearly as many specimens as that of all the other forms combined. The lateral stripe is always on the second and third scale rows, and the eye is large. The dorsal stripe usually occupies the median and halves of the adjacent rows, is frequently narrower and occasionally wanting. The dorsal scale formula is practically always (Gn 598 out of 600 specimens examined; see p. 181) 19-17; the supralabials nearly always 7, occasionally 8, rarely 6; the infralabials 10, occasionally 9 or 11. The oculars are usually 1-3, often 1-4, more rarely 1-2; the subcaudals 54 (female) to 84 (male); the ventrals 137 (female) to 167 (male). Tail length .192 to .262. The ground color above varies from light green, light greenish olive, or light olive brown to dark greenish olive, dark olive brown, brown, or black. The first row of scales is generally (but not always) much lighter than above, but still somewhat darker than the second and third rows. There are two rows of well-defined spots (rarely fused) on the skin between the lateral and dorsal stripes, and these usually cover more or less of the involved scales (exclusive of the « Thamnophis sirtalis (LINNAEUS), Syst. Nat., 10th ed., I, p. 222. Includes Coluber ordinatus LINNAEUS, Eutaenia sirtalis graminea Corr, E. sirtalis obscura Corx, E. sir- talis semifasciata Core, and E. sirtalis pallidula ALLEN. VARIATIONS OF GARTER-SNAKES. erie keels). The spots are not usually distinct upon the scales, however, owing to the darkness of the ground color. Interspaces on the skin white or bluish white, never red except occasionally immediately along the lateral stripe. Stripes greenish or bluish, usually tinged with yellow. Ventral surface greenish or bluish slate. Ventral spots usually well defined. Sides of head greenish, yellowish, or bluish; abials partly margined with black. Habits and habitat relations.—Even in the case of this well-known form but little has been recorded on the habitat relations. The expe- Fiag. 77 .— FORESTS IN THE PORCUPINE MOUNTAINS, MICHIGAN. THAMNOPHIS SIRTALIS IS OF GENERAL DISTRIBUTION IN SUCH FORESTED REGIONS IN EASTERN UNITED STATES. rience of the writer indicates that it is quite generally distributed in the Eastern forest region, for while it is found most commonly in the vicinity of water, it is not uncommon in the clearings, woods, and thickets on the neighboring hills (fig. 77). The food consists principally of frogs, toads, salamanders, earth- worms, and various insects. Whether or not it feeds to any great extent upon tadpoles and fish is undetermined. Garman (1892, 268) states that they eat these animals, and I have observed them to 178 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. capture fish in captivity, but since in the wild state they are not par- ticularly aquatic, the truth of the matter is probably that they cap- ture these forms when they encounter them in small pools, but that this iscomparatively seldom. The number and kinds of insects eaten is also a questionable point. It is true that many species are found in the stomachs examined, but, as Surface (1906, 149) says, many of these are ‘‘taken inside of the toads and other batrachians which the garter-snake had eaten.’’ However, both adults and young are very fond of earthworms. As other garter-snakes, sirtalis apparently does not refuse dead food. Mr. N. A. Wood, of the University of Michigan, reported to the writer, on May 18, 1907, that he saw a speci- men of this species swallowing a yellow warbler, which he had observed lying dead in the same place on the preceding day. In the latter part of October, 1907, the writer discovered an individual at Portage Lake, Washtenaw County, Michigan, busily engaged in an attempt to swal- low the dried remains of a large green frog (Rana clamitans). The breeding habits have been commented upon several times, but are as yet only incompletely known. In southern Michigan copula- tion takes place in April, and at this time it is reported on good authority that these snakes often collect in groups, probably owing to the procreative impulse. I have not witnessed this nor can I find any observations on the act of copulation. The latter I have seen but once, and then but imperfectly. It took place on April 21, 1906, between two specimens in captivity. The male in this case lay at full length beside the female, and evidently attempted to excite her by gently rubbing her neck with his snout. He finally threw a fold of his tail across hers, and turning his ventral surface against her side began spasmodic contractions of the abdominal muscles, which were continued from twenty to thirty minutes. Unfortunately the snakes were then disturbed and the observations ceased. They indicate, however, that there may be some interesting courtship reactions to be observed in these snakes. The period of parturition extends from the latter part of July to about the middle of September. Both of these dates are only approxi- mate, as definite observations are wanting. The number of young is very variable, the average range in number being probably about 10-30, while as many as 78 have been recorded in a single brood, which is not at all an unusual number, since parietalis may have, according to our observations, as many as-73. After birth the young remain for a short time about the mother, but this time is probably limited to a few hours at most. In captivity there is little tendency discernable to stay near the mother, and although we have several times seen a mother and her brood in a wild state, in every case noted, when the mother became alarmed, or for some other reason moved away, the young scattered in all directions, and it is improbable that they ever VARIATIONS OF GARTER-SNAKES. 179 came together again. The quickness and completeness with which the little snakes disappear when alarmed may partly explain the fable that this snake swallows its young. Ditmars (1907, 235-236) gives the following interesting account of the hibernating habits: The favorite situations in which to pass the cold months are in soft soil on a slope that faces the south. Here the reptiles burrow down a yard or more. Rocky situa- tions are often selected, and among the clefts and fissures, one opening into another, the snakes are enabled to retire to a considerable depth from the surface. It is in the fall that these snakes congregate in large numbers on ground that is suitable for the winter’s sleep. Here they sun themselves during the middle of the day, retiring into clefts and burrows during chilly autumn nights. As the nights become colder, their basking periods during the day are shortened, and finally, after the first severe frost, they remain below the ground for the winter. Instinct seemingly attracts them to these places of hibernation, for such spots are usually poor feeding grounds and have been devoid of snakes during the summer months. In spring, the breeding time, the reptiles remain in numbers until the weather has become well settled and the danger of needing good shelter from the cold spells has passed. Then they scatter into the ravines, the thickets, along streams and brooks, until the scene that has abounded with sinuous, crawling life is deserted. This account harmonizes very well with the writer’s observations in southern Michigan. In the latter region they are found in the autumn on sunny hillsides in the immediate neighborhood of holes, into which they hasten when alarmed, but that they dig these holes themselves yet remains to be proven,nor after the beginning of the period of hibernation do they necessarily ‘‘remain below the ground for the winter,” for if periods of marked moderation in the temperature occur they will ‘come out in December, January, or February. Thus, on January 22, 1906, which was a warm day (60° F.) in a period of very moderate temperature, a collector for the University of Michi- gan Museum reported seeing a large garter-snake near Grass Lake, Washtenaw County, Michigan, which was undoubtedly this species. Range.—This form is practically confined to the eastern wooded district of North America. Thus, its range on all four sides is rather definitely bounded—on the south by the Gulf, on the east by the Atlantic, on the north by its ability to endure low temperatures, and on the west by the margin of the prairie (fig. 78). Specimens have been examined from the following localities: Lac Aux Sables, Quebec; Auburn, Aroostook County, Androscoggin, Maine; Wellesley, Marthas Vineyard, Woods Hole, No Mans Land, Tuckermuck, Gloucester, Boston, Cohasset, Arlington, Cambridge, Massachusetts; Conanicut Island, Dutch Island, Chepachet Island, Newport, Rhode Island; Falmouth, Bartlett, New Hampshire; Mon- roe County, Delaware County, Philadelphia, Port Allegheny, Foxburg, Huntingdon County, Pennsylvania; Westport, Adirondack, Catskills, Syracuse, Tioga County, New York city, New York; Sussex County, New Jersey; Washington, District of Columbia; Centerville, Prince 180 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. George Island, Branchville, Maryland; Courtland, West Fork, Green- brier River, North Fork Potomac River, South Fork of Files Creek, Elk River (Cougar’s Mills), Beech Mountain, Cheatbridge, Huttonsville, Hinton, West Virginia; Page County, Fairfax County, Arlington, Suf- folk, Virginia; Glasgow, Kentucky; Tyree Springs, Union County, Tennessee; Kinston, Raleigh, Jackson, North Carolina; St. Simons Island, Georgia; Enterprise, Georgiana, Kissimmee River, Eustis, IETS USS OY ee pee AEs a #) Fic. 78.—DISTRIBUTION OF THAMNOPHIS SIRTALIS, AS INDICATED BY THE LOCALITY RECORDS. Arlington, Fort Bassenger, Clear Water, Tarpon Springs, Orlando, Florida; Eutaw, Mobile, Alabama; Kemper County, Mississippi; New Orleans, Louisiana; Fort Jessop, Hot Springs, Arkansas; Waco, Texas; St. Charles County, St. Louis, Jefferson County, Crawford County, Oregon County, Missouri; Aux Plains, West Northfield, Mount Carmel, Olney, Chicago, Paris, Rock Island, Lake County, Berwyn, Henderson County, Northfield, Ilinois; Lake Maxinkuckee, Vincennes, Bascom, Sims, Indianapolis, Winona Lake, Brookville, Montgomery County, VARIATIONS OF GARTER-SNAKES, 181 Indiana; Oberlin, Columbus, Sandusky, Montgomery County, Yellow Creek, Ohio; Washtenaw County, Grand Rapids, Grosse Isle, Living- ston County, Eaton County, Oakland County, Oceania County, Crawford County, Iosco County, Alma, Isle Royale, Porcupine Moun- tains, Bessemer, Houghton County, Baraga County, Marquette, Mich- igan; Thompson’s Lake, Racine, Lauderdale, Milton, Wisconsin; Lucknow, London, Wellington County, Gravenhurst, Ontario, Canada. - It occurs on the nearest of the outlying islands in the Atlantic Ocean and Gulf of Mexico. The most northern reliable records are Isle Royale, Michigan, and Gaspé, Quebee (Cox, 1899), but it doubtless ranges somewhat beyond this latitude. On the other hand, its western limits are neither well known nor definitely ascertainable, for in this direction it intergrades with parietalis, the ranges of the two forms being apparently conterminous. At present its western boundary may be considered as closely approximating that of the hard-wood forest. Without doubt typical specimens will be found to occur in Minnesota and Iowa, but this can not now be settled, inas- much as the principal character (red interspaces between the lateral spots) that distinguishes parvetalis from sirtalis disappears rapidly in preserved material. Variation.—N otwithstanding the wide range of sirtalis, the variation in scutellation is sight. As already stated, the dorsal scale formula is so constantly 19-17 that but two specimens out of some six hundred examined have any other. In these two specimens (one each from Fort Bassenger, No. 22696, and Georgiana, Florida, No. 14833, in the U. S. National Museum) the formula is 19-21-19-17. The labials are comparatively nearly as constant, the average formula for localities throughout the range being very close to 7/10. Occa- sionally specimens, regardless of their geographic position, exhibit 8 supralabials, and much more rarely 6, while contrary to this the variation in the infralabials 8, 9, or 11 apparently tend somewhat more strongly toward a lower number than 10. This is illustrated in the diagrams (figs. 79-80), which illustrate the conditions in 113 specimens from southeastern Michigan (Livingston, Washtenaw, Oakland, Eaton counties), and the variations in the material examined from every other locality is apparently very similar. The number of ventral plates varies from 137 to 167, and the average is rather constant throughout the range. I believe, how- ever, that there is a slight reduction in the mean number in speci- mens from extreme eastern United States. Thus, the diagram (fig. 81) shows that the average for specimens west of the Allecheny Mountains is very close to 154 (between 150-155), but in specimens from Maryland, Massachusetts, and Rhode Island the mean is be- tween 145-149. Based as they are on rather extensive series, I - 33553—Bull. 61—08——13 182 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. believe that these averages are quite representative of the actual conditions. The tail length and number of subcaudal plates also exhibit no noticeable geographic differences. The extremes are: Subcaudals 54-74 (female), 62-84 (male); average for localities with more than fifteen specimens, 64-70; tail length, .192 to .262; average of fifty- four specimens from Washtenaw County, Michigan, .221. Percentage of specimens. 98 eet feat ci ae Caakteea e STeMgd Soe © Supralabials. Fic. 79.—VARIATION IN THE NUMBER OF SUPRALABIALS IN 113 SPECIMENS OF THAMNOPHIS SIRTALIS ‘ FROM SOUTHEASTERN MICHIGAN. Similarly but few of the color phases have distinct geographic limits. About six of these forms have been described. (1) ordinatus Linneus.—This color phase is characterized only by the absence of the dorsal stripe, but even so, if the character were constant, some excuse might be had for considering it distinct from sirtalis. As it is, however, the same brood may contain individuals both of this and the normal phase, while among adults various intermediate stages in the distinctness of the dorsal stripe occur, so —ee ee VARIATIONS OF GARTER-SNAKES. 1838 that it is frequently impossible to refer a specimen definitely to either form. Nevertheless, while its range is not distinct from that Percentage of specimens. 98 18 .02 fet 9-9 8-8 Infralabials. ° Fic. 80.—VARIATION IN THE NUMBER OF INFRALABIALS IN 113 SPECIMENS OF THAMNOPHIS SIRTALIS FROM SOUTHEASTERN MICHIGAN. of sirtalis, it must be pointed out that the majority of the specimens examined hail from northeastern United States (east of Ohio and Illinois. Southern Northern Indiana. Ohio. Florida. Maryland. Massa- Rhode Michigan. Michigan. chusetts. Island Fig. 81.—DIAGRAM SHOWING THE VARIATION IN THE NUMBER OF VENTRAL SCUTES IN THAMNOPHIS SIRTALIS. north of North Carolina), those from outside of this region being but scattered records. . 184 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. (2) graminea Cope.—Typical specimens of this phase may be defined as follows: Above light green usually clouded with yellow. Lateral spots not evidenced on scales, except occasionally anteriorly, but present on the skin. Interspaces on the skin bright yellow. Belly light green, often tinged with yellow. Dorsal and lateral stripes absent. This form is very similar to the ordinatus phase except that the ground color above is lighter, the dark spots are mostly absent from the scales, and the lateral stripe is not discernable. But, as in ordinatus, all intermediates are found between this color phase and that of typical sirtalis. In specimens in which the ground color is somewhat darker the dorsal stripe is indicated by a lighter shade on more or less of the three median dorsal rows, the lateral by a similar shade on the second and third rows. This color phase is very striking, and to one not familiar with the variability of the form it would seem quite desirable to recognize it as a distinct form, the more so as it is apparently common in many localities. From the proportion of specimens in the collections it seems to occur principally in eastern Ohio, becoming rarer in the western part of the State, but extending into eastern Indiana. (3) pallidula Allen.—But little attention has been given to this phase since its original description. It may be described as follows: Ground color above dark. Spots obscured on scales. Stripes dull greenish or bluish, the dorsal more or less obscure. This form differs from ordinatus only in the obscurity of the lateral spots and the dark- ness (often amounting to partial obscurity) of the stripes. It has been recorded from the vicinity of Intervale and Bartlett, New Hampshire, but the University of Michigan Museum expedition secured specimens on Isle Royale, Lake Superior, that are indistin- guishable from the topotypes of pallidula in the National Museum. (4) semifasciata Cope.—The character upon which this form has been distinguished is very slight and strikingly like the one upon which the form 7. eques collaris was based. In the specimens referred to semifasciata the first two to five spots of the two rows between the lateral and dorsal stripes are opposite and confluent transversely, forming blotches that may or may not cross the lateral stripe. This coloration occurs very frequently in specimens in and immediately about the ‘prairie peninsula,” in Indiana, Illinois, southern Michigan, and southern Wisconsin, and apparently much less commonly to the east and south of this region. But not all of the specimens from this region exhibit this tendency, toward the fusion of the lateral spots into transverse bands anteriorly, in the same degree, Many individuals are found which either have but one or two or even no blotches, and it is evident that the character is far from being a constant one. It is significant that the region where this color phase is best developed is close to the range of parietalis, which often exhibits the same phenomenon. VARIATIONS OF GARTER-SNAKES. 185 (5) obscura.—This so-called variety of various authors has had an interesting history. Defined by Cope as differing from sirtalis in the fusion of the lateral spots, it has been made by various writers to embrace slightly anomalous specimens of sirtalis, parietalis, and butlerr. .The original specimens from Westport, New York, are sirtalis, and they are peculiar only in that the ground color above is so dark as to obscure the spots on the scales (by no means an uncom- mon occurrence). The ‘obscura”’ specimens of Morse which are not referable to butleri are also of this description. Cope (1900, 1074) further describes specimens from Lac Qui Parle, Minnesota, as follows: “The fusion is complete as to the superior row of spots, but the infe- rior may be seen faintly outlined on stretching the skin, as in some dark forms of F. s. parietalis.”” This would seem to denote that the ‘spots on the other specimens referred to obscura by Cope (those from Westport, New York) were entirely fused on the skin; but this is not true, for on stretching the skin of the latter the spots are seen to be perfectly distinct. As a matter of fact the Minnesota specimens referred to obscura by Cope are parietalis (in which the upper row of spots is normally fused), the New York specimens of this writer and the Ohio specimens (in part) of Morse being rather dark specimens of sirtalis. I have already alluded to the fact (see p. 88) that both Morse and Cope have referred specimens of butleri to obscura. This was made possible by the insufficiency of the original description of the latter, which gives the impression that the spots are fused on the skin in typical specimens, which is the case to a considerable extent in butleri. (6) In many specimens from Indiana, Ohio, southern Michigan, and western Pennsylvania the skin between the first and third or fourth scale rows is more or less red. On the basis of such specimens it has been asserted that parietalis occurs in the above named States. These specimens are typical sirtalis in every other character, while the red when present is always in such a small amount that the speci- mens need never be confused with parietalis. While in my judgment it is impossible to recognize any of the above forms as distinct, I believe that the first chnee (ordinatus, grami- nea, and pallidula) are significant in that they represent the same tendency in the form, i. e., toward an increased darkness in color. I believe that the increasing darkness of the stripes (which leads to their obscurity) is an evidence of such a tendency. This explains the fact that the extremes in this direction (those in which the stripes are obscured and which have been described as distinct) have no distinct geographic range, but are liable to crop up in any consider- able collection of specimens from almost any locality in the range of the form. Nevertheless it seems to be evident also that the tendency is much more pronounced in the north and northeastern than in south and southeastern parts of the range, although a considerably larger 186 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. number of specimens is needed from the latter region before this point can be definitely demonstrated. Affinities —The only apparent near relative of sirtalis is parietalis. I have already discussed this relationship. It should be pointed out, however, that the fact that typical specimens of sirtalis, from Indiana to western Pennsylvania, have frequently a small amount of red pigment on the sides, also points toward a relationship between these two forms. Brown’s (1904, 470-471) ingenious suggestion that sirtalis is related to radix through butleri has little to recommend it, for from my point of view there is no evidence from the scutellation that sirtalis is at all nearly related to butlerr, although the general similarity in the scutellation and the intermediate position of the lat- eral stripe in the latter has been occasionally seized upon as indicative | of such an affinity, while there is no evidence either in the position of the lateral stripe, scutellation, or geographic distribution that it is anything but distantly related to radix. To establish a relationship with the forms of the Radix group one must entirely ignore the evi- concinnus —_———— parietalis -----_ sirtalis eques sumichrasti Fic. 82.—PHYLOGENETIC DEVELOPMENT GF THE SIRTALIS GROUP. dence of the lateral stripe and the geographic probabilities, as sirtalis entirely overlaps the range of butleri, and in part the range of radia, without showing the slightest modification in any character that would point toward any close relationship with either form. CONCLUSION. Of the four groups into which I believe the genus Thamnophis is differentiated, the Sirtalis group is without doubt the least diver- sified. This is shown by the fact that most herpetologists have been willing to give to its members but subspecific rank, owing to their evident intergrading, and by the further fact that variation, geo- graphic or otherwise, is very slight. If we grant the unity of the croup as here described (fig. 82) the geographic differences may be outlined as follows: 1. The dorsal scale rows are 19-17 in most forms, the only departure from this formula being in southern Mexico, where it becomes 17-15. ancpst leit si A Sill tl al de vee eee es VARIATIONS OF GARTER-SNAKES. 187 2. The supralabials are normally 8 in the two forms in Mexico and 7 in the others. The infralabials are normally about 10 in all forms. 3. The ventrals, subcaudals, and tail length are, so far as known at present, practically the same throughout the entire range of the group. Thus the geographic variations in the scutellation consist of a maxi- mum of 19-17 dorsals and 8 supralabials in the form whose principal distribution is on the Mexican plateau (eques), and of a reduction in the dorsal scale formula in the forms to the southward and in the number of supralabials in those to the northward of the range of this form. It only remains to be pointed out that the area of principal reduction between the forms with a different scutellation is small. DISCUSSION OF ORIGINS. In the general discussion of the characters and variation in the genus, as well as in the detailed discussion of the various forms, I hold to have established two fundamental facts as follows: (1) That the genus Thamnophis is naturally divided into four groups of nearly equal importance, the component forms of which are directly related; (2) that each of these groups is represented by a form in the northern part of Mexico and in southwestern United States which has the maximum scutellation for the group. To what extent do these facts throw light upon the question of origins? Darwin pointed out in his Origin of Species that on the principle of descent with modification the only rational view is that the several species of a genus have been derived from a common progenitor and have undergone modification in the course of dispersal. We are then confronted with the problem of locating the center of origin of a genus before we can intelligently search for laws governing the modifications which have resulted in the component forms as we find them at the time of our study. Various criteria have been formulated toward this end.* At least one of these can not be used in this instance, as it is based on peculiar conditions existing in one group of animals: it is the direction of the annual migration routes of birds (Palmén). Of the others I believe that Adams’s sixth criterion is of the most general and easy appli- cation as well as logically (assuming, of course, the theory of descent and single centers of origin) the most reliable. This rule is that the different lines of dispersal in any genus radiate from, or, conversely, converge toward, the geographical center of origin. As already men- tioned, I believe that this rule follows logically from the theory of descent formulated by Darwin, so that if only the lines of dispersal can be determined in any group in which there are more than one, it should be relatively easy to fix upon the center of origin of that genus. In discovering that different nearly equal groups of related «See Adams (1902 6). 188 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. species occur in the garter-snakes, I have proof that different lines of dispersal have been followed; and it will furthermore be recalled that, as I have defined these groups, they are each composed of forms which are distributed linearly. When I find, then, that nowhere else in the range are all of the groups represented, I am justified in concluding that the genus has had its origin in a general way in northern Mexico. As a corollary to this criterion is Adams’s first test, namely, that the center of origin is the location of the greatest differentiation of a type. If the lines of dispersal converge toward or radiate from the center, in this general region and here only, it should be possible to find representatives of each group or lines of directly related forms. This is exactly what occurs in the genus Thamnophis. In no region except Mexico can there be found representatives of all of the four groups of garter-snakes, while at least to the northward representa- tives of but one or two groups are found toward the outlying parts of the range. The application of either of the above tests, however, can not, in my opinion, be trusted to indicate more than the general region in which the center of origin of the garter-snakes is located, for in a constricted area like Mexico and Central America the groups differ- entiated in one part must of necessity remain together geograph- ically until they emerge into a more extensive land drea, such as North and South America. This is exactly what occurs, for the four groups occur throughout Mexico, although, as will be shown later, they prob- ably became differentiated in a particular part of this general region. Another test that has been suggested is the one laid down by Allen that synthetic forms will be found at the center of origin. It is one that naturally follows an acceptance of the theory of single centers of origin, and is closely bound up with the first two criteria cited— the convergence of lines of dispersal and the location of the greatest differentiation of a type. At the center of origin of the genus of garter-snakes, then, theoretically all four groups should be repre- sented, and the representatives of these groups should be more closely related than any other representatives of the groups. This point is difficult to test, owing to the fact that the different groups as they diverge genetically and often geographically may converge in appearance. This reduplication of forms in widely separated areas occurs to such an extent in the garter-snakes as to make it often difficult or even impossible to determine specimens without a knowl- edge of the locality in which they were taken. This parallelism is brought about by the reduction in size which takes place as the groups depart from the center, as this reduction in size is accom- panied by a reduction in scutellation. Thus every group has forms which are practically identical in the number of scales in each series. ti psi ty OEY Map ett Oe! Madge 4 VARIATIONS OF GARTER-SNAKES. 189 This is best shown in the end forms. Thus in butler: (Indiana, Ohio, etc.), ordinoides (coast of Washington, Oregon, and northern Cali- fornia), scalaris (Veracruz), and swmichrasti (Veracruz) the scutel- lation is often identical, while the coloration in but one of these, scalaris, can be relied upon as a differential character. As stated above, the similarity of such forms often makes it difficult to dis- tinguish between parallel and synthetic types, but an exact knowl- edge of the value of differential characters and of geographic distri- bution and trends of variation will allow the distinction to be made. While it is thus impossible to accept similarity in scutellation be- tween the forms of different groups as always indicative of nearness of relationship, it is significant that where the groups come together geographically their representatives also come to resemble each other closely in scutellation. Thus, in megalops, angustirostris, eques, and proximus the labials are practically the same in number, and the number of ventral scutes is as nearly the same as can be expected for so variable a series. It is true that the usual dorsal scale formula in megalops and angustirostris is 21-19-17, while for proximus and eques it is 19-17, but it must be remembered that it is nowhere higher than 19-17 in the respective groups to which eques and proximus belong, so that these forms resemble megalops and angustirostris more closely than the two latter are resembled by any other forms of the Sauritus and Sirtalis groups. While, therefore, similarity in scutellation does not necessarily in itself (especially at the limits of the range of the groups) imply close rela- tionship, if we grant the evidence of the other criteria that the center of dispersal is in Mexico, there is no objection from this standpoint, for angustirostris, megalops, eques, and proximus give every indica- tion of being more closely related to each other than any other four forms in the different groups. If this is true it restricts the center of origin of the genus to northern Mexico. Two other criteria of Adams may be considered together, i. e., “continuity and direction of individual variation or modification radiating from the center of origin along the highways of dispersal” and ‘direction indicated by biogeographical affinities.” Evidently, it seems to us, both of these criteria are of importance in any investiga- tion of this kind, for, while they are not concerned primarily with the conditions at the center, they are supreme tests of the genuineness of the lines of dispersal, which must be determined before the center can be located. The tests can be reformulated as follows: Are the variations in the forms along any supposed lines of dispersal con- tinuous, and is it probable geographically that the forms are related in the supposed manner? For instance, I might divide the garter- snake into several groups on the basis of similarity in the number of scale rows, but if I then applied these tests we would find that 190 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. not only would the forms of each group fail to intergrade with each other in their variations, but that it would be extremely improbable, in view of their geographic location, that they were at all closely related. As it is, each group is formed of a line of directly related forms, the extremes of which are very distinct, but those which are geo- graphically nearest together are but slightly different from each other. As previously stated, I believe the tests used above are valid, and when applied show the utter untenableness of the hypothesis put forward by Cope and Brown that the center of dispersal is in southeastern United States and that sirtalis is the ancestral form. From the latter standpoint the groups which we have distinguished in the genus must be ignored and the forms connected in a manner entirely antagonistic to the criterion that the lines of dispersal shall be indicated by continuity in variation and be in harmony with geographic probabilities. Note in the diagram of affinities given by Brown (1904, 471) that the lines of genetically related forms fullfil the requirements of neither of the above tests, and that, admitting the validity of our own groups, we can get, at most, but three lines radiating from eastern North America, and one of these does not reach the southeastern States! If the criteria discussed above are valid I believe that they indicate unquestionably that the center of origin is in northern Mexico, and I may test the other criteria that have been formulated by different writers. Two other criteria of Adams may be considered together. They are: That the center of origin will be marked by a ‘‘dominance or great abundance of individuals” and by a minimum ‘‘dependence upon a restricted habitat.’ Neither of these postulates is true in . this genus, for, although much yet remains to be discovered on the habitat relations and abundance of the different forms, garter-snakes are not noticeably abundant but rather rare in the desert and coastal regions of northern Mexico and southern New Mexico and Arizona, while in many places in North America they are almost increditably numerous. Again, on the plateau of Mexico and in southern New Mexico and Arizona, which forms the home of three of the groups, the conditions are very arid, and these snakes are evidently confined principally to the immediate vicinity of mountain streams and similar environments where water is most abundant. On the other hand, in the forested regions of North America, notably in eastern United States, the forms, although showing habitat preferences, are of quite general distribution. This is exactly what one might expect in view of the preference shown by these snakes for a moist habitat. It has been objected that this habitat preference argues against the con- clusion that the genus originated in northern Mexico. I can not see the force of this argument, for it seems to me that partially aquatic @See Van Denburgh (1897, 212) and Ruthven, p. 73, in this paper VARIATIONS OF GARTER-SNAKES. 191 forms may originate in favorable environments in arid regions, just as strictly aquatic forms may do so. I believe, therefore, that, in cases of this kind at least, these two criteria (abundance and least restriction to a particular habitat) are not valid, for a group may originate in a region where the environmental conditions to which it is adapted are more restricted than elsewhere. The fifth criterion of Adams, ‘‘Location of greatest productiveness and its greatest stability” (Hyde, 1898, p. 575), is in the main inap- plicable in this study, owing to our almost total ignorance of the breeding habits of the various forms. It seems worth recording, however, that the most dwarfed forms (those with the smallest scutellation) have apparently. a fewer number of young in a brood than do the larger forms, as might be expected. For example, butlert generally has about a dozen young in a brood, while radix may have forty or fifty. As the relative productiveness thus seems to be correlated with size in this genus, the location of the area of greatest productiveness would be near the center of origin, but even if this should be found to be the case it would prove only that the test can be used in genera in which the forms are the result of dwarfing. However, as said above, so little is known of the breeding habits of these snakes, that it is hazardous to attempt any general conclusions at this time. Another criterion that is quite easy to apply is the one formulated by Allen in 1877 (1877, 378): ‘‘The representatives of a given species increase in size toward its hypothetical center of distribution, which is in most cases doubtless also its original center of dispersal.’ In snakes it is doubtful if we have any such a thing as ‘“‘adult size” in the sense employed in birds and mammals, so that, without some method of indicating dwarfing, this criterion is valueless. Fortu- nately we believe that differences in the number of scales in the dorsal, labial, and ventral series is, as outlined in the chapter on variation, an index to the relative size in these snakes, and, if so, we have in these characters a ready means of determining the relative dwarfing of the different forms, even if young specimens alone are available for study. Applying this criterion to the conditions outlined in the description of the different forms the results are startling, for in every group the maximum scuiellation is found in the form which inhabits northern Mexico, while from this region the scutellation decreases progressively im the different forms, so that the minimum is only found in the forms which constitute the extreme ends of the series, and are thus genetically and geographically most distant from the form with the maximum number (at the center). In other words, the forms of the same group are progress- wely more dwarfed away from northern Mexico. [This does not include those forms of the Sirtalis group (concinnus and sirtalis), which are 192 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. based only on color characters.] The results are exactly those which we should expect if the criterion is a valid one, and the center of origin is in the northern Mexico, as indicated by the convergence of the lines of dispersal, but it must be recognized that while these facts may be in harmony with the theory, they can not be considered as proving it until the cause of dwarfing is known. METHOD OF EVOLUTION OF THE FORMS. It is not the purpose of this paper to discuss the causes of evolution in the garter-snakes, for the study of distribution and variation alone can not encroach upon the field of experimental morphology in deter- mining the causes and effects of the processes which effect the animal form. It is the province of such studies, however, not only to point out probable origins and relationships, but also to search for the methods of evolution, so that the experimentor may work intelli- gently in his search for the processes which cause the production of one species from another. If the range of the forms in the different groups of garter-snakes be carefully examined it will be found (1) that the different forms of the same group are found in different geographical regions, characterized by different environmental conditions; (2) that the area along the common boundary of two forms of the same group, where transition in characters takes place, is relative narrow. These conditions are exactly in harmony with Stone’s (1903, 659) statement for terrestrial vertebrates that ‘“we never find two geo- eraphic races or subspecies of the same form occurring together, except during times of migration,” and Jordan’s (1905, 547) statement that ‘‘given any species in any region, the nearest related species is not likely to be found in the same region nor in a remote region, but in a neighboring district separated from the first by a barrier of some sort.’’ Steere (1894, 419) evidently had the the same point in mind when he formulated the law ‘‘the genus is represented by only a single species in one place,’”’ but he expressed it too generally, for as a matter of fact, as shown in this genus, two or more species may occur together in the same environment, but the direct relative of any species is to be found not in the same but in a neighboring envi- ronment. In the light of the results of this investigation, as well as the known instances in which different species occur in the same geographic area but in different environments, it seems expedient to reformulate the law to read as follows: Directly related forms on any line of descent generally occupy neighboring environments. Although expressed in different terms, this is exactly the same conclusion arrived at by Ortmann (1906, 512) in his study of the crawfishes of Pennsylvania. At the present time it seems to be becoming recognized that the above law is of wide application among the higher orders of verte- VARIATIONS OF GARTER-SNAKES. 198 brates. However, any attempt to determine from geographic and variation data that internal factors, the direct influence of environ- mental conditions, or in fact any other cause is responsible for this close correlation of particular forms with particular environments must fail to be conclusive, for experimental work alone can definitely reveal the influence of the environment upon the dwarfing and scu- tellation in these snakes. In the case of the garter-snakes, however, it shouldbe noted: 1. That most of the forms are the result of dwarfing. 2. That the amount of dwarfing does not seem to be directly as- sociated with the nature of the environment, for the form inhabiting a particular region is only slightly different from its nearest neighbor in the same group, while forms of widely different scutellation may inhabit the same region. Thus the conditions which apparently deter- mine the scutellation of any form is the scutellation of its immediate progenitor, and the dwarfing which it has itself undergone. While therefore—although it is without the province of the present work to seek to determine just what factors in the environment make it unfavorable for these snakes—lI believe that I am justified in concluding that the dwarfing is associated in some way with the environment; it is as if the different environmental complexes en- countered by the groups as they radiated from the center of dispersal were generally unfavorable, and the most tenable hypothesis is that the changed conditions of environment acted as a stimulus (unfavor- able to growth) upon these snakes (Tower, 1906, 294-296). If this is true, isolation while attending the production of the forms in each group has had no other apparent effect than that of keeping them distinct as they were formed. Allen (1876, 310) concludes from such evidence that the environ- mental conditions at the center are the most favorable for the existence of any group, and Adams (1904, 210) that the center of dispersal corresponds to a center of optimum vital conditions. It follows from Adams’ standpoint that as a form enters other regions in which there are any changes in the optimum conditions these changes may be considered as unfavorable, whether they are in the way of an increase or decrease, since the zero point of a stimulus corresponds to the vital optimum and is situated between the minimum and maximum, both of which result in death. However this may be, the following facts will stand: (1) That the maximum scutellation and size in the genus Thamnophis occurs at the center of dispersal, and the forms that have been produced in thé history of its migration have been formed principally by dwarfing and by a reduction in scutellation; (2) that the variation in the number of scales in the different series is definite and not promiscuous, and is correlated in a remarkable degree with changes in the environment. 194 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. The development of the different groups has thus been orthogenetic. This is nowhere better shown than in such forms as radix, where, as one passes from the western to the eastern parts of the range, not only the average but also the maximum and minimum num- bers of plates in the different series decrease steadily, until the range of butleri is reached; and that this is in some way associated with the environment is shown in the numerous instances in which the transition from the scutellation of one form to that of another takes place rapidly, and along the boundary of different sets of envir- onmental conditions. Even in the latter instances the variations are evidently orthogenetic, as shown by the fact that geographically there are marked differences in the maximum, minimum, and aver- age number of scales in the different series. I have failed to find the slightest evidence that forms have been produced by mutations in this genus. From these facts, then, it seems to me that the most tenable hypothesis of the evolution of the genus Thamnophis is that it orig- inated and became differentiated inte four main groups in northern Mexico. From this region the groups radiated in all directions, but principally to the northward, and wherever they entered different regions the changed environmental conditions acted as an unfavor- able stimulus, which retarded growth, and differentiated the groups into dwarfed forms. If this explanation is the true one, the influence of the environment without doubt affects the young snake before birth, since the scutellation, at least as regards the number of scales in the dorsal, labial, ventral, and subcaudal series, is determined before the young are born. THE GARTER-SNAKES AS MATERIAL FOR EXPERIMENTAL INVESTIGATION. So far as I know, snakes have rarely been used for experimental purposes. They are unfitted for this both by the fact that apparently but one brood is raised each year and that they are difficult to keep in captivity. The garter-snakes, however, are free from both of these objections, since they are hardy in captivity, and while raising but one brood a year are very prolific. For these reasons they should present: very good material for the investigator, the more so as they may be had in numbers almost anywhere in the United States. It should be borne in mind, however, that when in captivity they must be kept, as nearly as possible, in natural conditions. Undoubtedly the problems that should first be attacked are the inheritability of the scale characters and the influence of different factors (inbreeding, unfavorable conditions of food and temperature) in producing dwarfing. The latter problem may be attacked in two ways: An attempt may be made to cause an increase in size and scutellation in dwarfed forms, such as butleri and ordinoides, or to ee a ee ee ee eee ee eee — Ss ee _— 2 Stee ogy VARIATIONS OF GARTER-SNAKES. 195 produce dwarfing in the larger forms, such as radix,.elegans, and megalops. In selecting material for such work, however, the appa- ently more variable forms of the Radix and Elegans groups should be chosen in preference to the less variable ones of the Sirtalis or Sauritus groups. Properly handled, these snakes should furnish the best material not only for the investigation of the problems peculiar to the order, but also for the study of general problems of heredity and evolution. BIBLIOGRAPHY. 1890. Assport, ©. C. A Naturalist’s Rambles about Home. New York. 1902a. Apams, ©. ©. Postglacial Origin and Migration of the Life of Northeastern United States. Jour. Geog., I, pp. 303-310; pp. 352-357. 1902b. : . Southeastern United States as a Center of Geographical Dis- tribution of Flora and Fauna. Biol. Bull., III, pp. 115-181. 1904. Y . On the Analogy between the Departure from Optimum Vital Conditions and Departures from Geographic Life Centers. Sci., N. S., XIX, pp. 210-211. 1876. Auten, J. A. Geographical Variation among North American Mammals, especially in Respect to Size. Bull. Geol. and Geog. Surv. Terr., II, No. 4, pp. 309-344. 1877. ; . The Influence of Physical Conditions in the Genesis of Species. Radical Review, I, pp. 108-140. Reprinted Rept. Smith. Inst., 1905, pp. 370-402. 1899. AttEN, G. M. Notes on the Reptiles and Amphibians of Intervale, New Hampshire. Proc. Bost. Soc. Nat. Hist., XXIX, pp. 63-75. 1901. Atkinson, D. A. The Reptiles of Allegheny County, Pennsylvania. Annals Carnegie Museum, I, pp. 145-157. 1905. Bamny, VerRNoN. Biological Survey of Texas. North Amer. Fauna, No. 25. 1853. Barry, S. F. and Grrarp, CHARLES. Catalogue of North American Reptiles, Pt. 1. 1895. Baker, F.C. A Naturalist in Mexico. Chicago. 1893. Boutencer, G. A. Catalogue of the Snakesin the British Museum,I. London. 1896. : . Catalogue of the Snakes in the British-Museum, IIT. London. 1904. Branson, E. B. Snakes of Kansas. The Kansas Univ. Sci. Bull., XIII, pp. 353-430. 1901. Bray, W. L. The Ecological Relations of the Vegetation of Western Texas. Bot. Gaz., XXXII, pp. 99-123; pp. 195-217; pp. 262-291. 1904. , . Forest Resources of Texas. U.S. Dept. Agric. Bur. Forestry, Bull. No. 47. 1882. Brous, H. Amer. Natur., XVI, pp. 564-567. 1889. Brown, A. E. Description of a New Species of Eutaenia. Proc. Acad. Nat. Sci. Phila., 1889, pp. 421-422. 1901. : . A Review of the Genera and Species of American Snakes, North of Mexico. Proc. Acad. Nat. Sci. Phila., 1901, pp. 10-110. 1903. : . The Variation of Eutaenia in the Pacific Subregion. Proc. Acad. Nat. Sci. Phila., 1903, pp. 286-297. Post-Glacial Nearctic Centres of Dispersal for Reptiles. Proc. 1904. ; Acad. Nat. Sci. Phila., 1904, pp. 464-474, 196 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. 1903. Crank, H. L. The Short-Mouthed Snake (Hutainia brachystoma Cope) in Southern Michigan. Proc. Biol. Soc. Wash., XVI, pp. 83-87. 1860. Cooper, J. G. Report upon the Reptiles Collected on the Survey. Rept. Pac. R. R. Surv., XII, Book II, pp. 292-306. 1862. Corr, E. D. Contributions to the Ophiology of Lower California, Mexico, and Central America. Proc. Acad. Nat. Sci. Phila., 1861, pp. 292-306. 1866. 5 . On the Reptilia and Batrachia of the Sonoran Province of the Nearctic Region. Proc. Acad. Nat. Sci. Phila., 1866, pp. 300-314. 1880. : . On the Zoological Position of Texas. Bull. U. S. Nat. Mus., XVII. 1883a. ; . Notes on the Geographical Distribution of Batrachia and Reptilia in Western North America. Froc. Acad. Nat. Sci. Phila., 1883, pp. 10-35. ; 18830. A New Snake from New Mexico. Amer. Nat., 1883, pp. 1300-1301. 1885a. j . A Contribution to the Herpetology of Mexico. Proc. Amer. Phil. Soc., XXII, pp. 379-390. 1885b. ; . Twelfth Contribution to the Herpetology of Tropical America. Proc. Amer. Phil. Soc., XXII, pp. 167-194. 1886. ; . Thirteenth Contribution to the Herpetology of Tropical America. Proc. Amer. Phil. Soc., 1885, pp. 271-287. 1888. : On the Eutzeniz of Southeastern Indiana. Proc. U. S. Nat. Mus., XI, pp. 399-461. 1892. ‘ . A Critical Review of the Characters and Variations of the Snakes of North America. Proc. U. 8. Nat. Mus., XIV, pp. 589-694. 1900. F -. The Crocodilians, Lizards, and Snakes of North America. Rept. U. S. Nat. Mus., 1898, pp. 153-1270. 1875. Cours, E. Synopsis of the Reptiles and Batrachians of Arizona. Wheeler’s Surv. West of 100th Meridian, V, pp. 585-633. 1878. Cours, E.and Yarrow, H.C. Noteson the Herpetology of Dakota and Montana. Bull. U. 8. Geol. Surv. Terr., IV, pp. 259-291. 1899. Cox, Pump. Freshwater Fishes and Batrachia of the Peninsula of Gaspé, P. Q., and their Distribution in the Maritime Provinces of Canada. Trans. Roy. Soc. Can., (2) V, Art. IV, pp. 141-164. 1883. Davis, N.S., Jr., and Rice, F. L. List of Batrachia and Reptilia of Illinois. Trans. Chicago Acad. Sci., II, pp. 25-32. 1907. Dirmars, R. L. The Reptile Book. New York. 1901. Ecxret, E. The Snakes of New York: An Annotated Check List. Amer. Natur., XX XV, pp. 151-155. 1880. Exnicorr, E. L. American Naturalist, XIV, p. 206. 1843. Firzincrer, L. 1. Systema Reptilium. Vienna. 1883. GARMAN, S. The Reptiles and Batrachians of North America. Mem. Mus. Comp. Zool., VIII, No. 3. , H. A Synopsis of the Reptiles and Amphibians of Illinois. Bull. Ill. State Lab. Nat. Hist., III, pp. 215-388. 1894. Giinruer, A. C. L.G. Reptilia and Batrachia. Biologia Centrali-Americana. 1892. 1907. GRINNELL, J. and GrinnNELL, H.W. Reptiles of Los Angeles County, California. Throop Inst. Bull., XX XV. 1852. HattoweELtt, Epwarp. Descriptions of New Species of Reptiles from Oregon. Proc. Acad. Nat. Sci. Phila., 1852, pp. 182-183. VARIATIONS OF GARTER-SNAKES. 197 1881. Hay, O. P. Euteenia radix in Indiana, American Naturalist, XV, p. 738. 1887. ; . A Preliminary Catalogue of the Amphibia and Reptilia of the State of Indiana. Jour. Cinn. Soc. Nat. Hist., X, pp. 59-69. 1892a. On the Breeding Habits, Eggs, and Young of Certain Snakes. Proc. U. S. Nat. Mus., XV, pp. 385-397. 1892b. The Batrachians and Reptiles of the State of Indiana. 17th Ann. Rept. Indiana Dept. Geol. and Nat. Resources, pp. 409-602. 1900. Hemprin, A., in Mux, H. R. The International Geography, pp. 774-781. New York. 1900. Hitt, R. T. Physical Geography of the Texas Region. Topographic Atlas, U.S. Geol. Surv. 1898. Hyp, Jonn. Variations in the Rate of Agricultural Production and One of its Causes. Science, N. 8., VIII, pp. 575-576. 1896. Jackson, R. T. and Jaccrr, T. A. Studies of Melonites multiperus. Bull. Geol. Soc. Amer., VII, pp. 135-170. 1905. Jorpan, D.S. The Origin of Species through Isolation. Science, N. S., XXII, pp. 545-562. 1860. Kennicott, Ropert. Descriptions of New Species of North American Ser- pents in the Museum of the Smithsonian Institution, Washington. Proc. Acad. Nat. Sci. Phila., 1860, pp. 328-338. 1758. LinNE, CARL von. Systema Natura, 10th edition. Upsalia, Sweden. aA ial 1766. i . Systema Natura, 12th edition. Upsalia, Sweden. , 1894. LoENNBERG, Ernar. Notes on the Reptiles and Batrachians Collected in Florida in 1892 and 1893. Proc. U. S. Nat. Mus., XVII, pp. 317-339. 1899. Merx, S. E. Notes on a Collection of Cold-blooded Vertebrates from the Olympic Mountains. Field Mus. Nat. Hist., Zool. Ser., I, pp. 225-236. 1896. Metpoua, R. The Speculative Method in Entomology. Ann. Meet. Entomol. Soc. Lond., Jan. 15, 1896. 1891. Merriam, C. Harr. Results of a Biological Reconnoissance of South-central Idaho. North Amer. Fauna, No. 5. 1904. Morsr, Max. Batrachians and Reptiles of Ohio. Proc. Ohio State Acad. Sci., IV, pp. 91-144. 1906. Ortmann, A. E. The Crawfishes of the State of Pennsylvania. Memoirs Carnegie Museum, II, pp. 343-524. 1900. Pounp, R. and CLements, F. E. The Phytogeography of Nebraska. Lincoln, Nebraska. 1895. Reppick,G. Snakesof Turkey Lake. Proc. Ind. Acad. Sci., 1895, pp. 261-262. 1904. Ruruven, A. G. Butler’s Garter Snake. Biol. Bull., VII, pp. 289-299. 1906. , . The Cold-blooded Vertebrates of the Porcupine Mountains and Isle Royale, Michigan. Rept. Geol. Surv. Mich., 1905, pp. 107-112. 1906a. RuTHveNn, A. G., in Cor, L. J. and Barsour, T. Reptilia, Amphibia, and Pisces. (Vertebrata from Yucatan.) Bull. Mus. Comp. Zool., L, pp. 146-159. 1907. ; . A Collection of Reptiles and Amphibians from Southern New Mexico and Arizona. Bull. Amer. Mus. Nat. Hist., XXIII, pp. 483-604. 1882. Scuneck, J. A Prolific Garter Snake. Amer. Natur., XVI, p. 1008. 1882. Smirn, W. H. Report on the Reptiles and Amphibians of Ohio. Geol. Surv. Ohio, IV, pp. 633-734. 1894. Steere, J. B. On the Distribution of Genera and Species of Nonmigratory Land-Birds in the Philippines. Ibis, 6th Ser., VI, pp. 411-420. 33553—Bull. 61—08——14 198 1894. 1903. 1906. 1906. 1892. 1906. 1905. 1896. 1897. 1875. 1883. BULLETIN 61, UNITED STATES NATIONAL MUSEUM. STEJNEGER, LEonNHARD. Notes on Butler’s Garter-Snake. Proc. U. S. Nat. . Mus., XVII, pp. 593-594. Stone, Wirmer. Racial Variation in Plants and Animals, with Special Ref- — erence to the Violets of Philadelphia and Vicinity. Proc. Acad. Nat. Sci. Phila., 1903, pp. 656-699. ; . Notes on Reptiles and Batrachians of Pennsylvania, New Jer- sey, and Delaware. Amer. Natur., XL, pp. 159-170. Surrace, H. A. The Serpents of Pennsylvania. Bull. Div. Zool., Penn. State Dept. Agri., IV, Nos. 4 and 5. Taytor, W. E. The Ophidia of Nebraska. Ann. Rept. Neb. State Board Agri., 1891, pp. 310-357. Tower, W. L. An Investigation of Evolution in Chrysomelid Beetles of the Genus Leptinotarsa. Carnegie Inst., Washington, Pub. No. 48. TRANSEAU, E.N. Forest Centers of Eastern America. Amer. Natur., XX XIX, pp. 875-889. Tutt, J. W. Philosophical Aspects of Entomology. Proc. City of Lond. Entomol. and Nat. Hist. Soc., Dec., 1896. Van Densurcu, JouHn. The Reptiles of the Pacific Coast and Great Basin. Occ. Papers Calif. Acad. Sci. V. Yarrow, H. ©. Report upon the Collections of Batrachians and Reptiles. Wheeler’s Surv. West of 100th Merid., V, pp. 509-584. i . Descriptions of New Species of Reptiles in the United States National Museum. Proc. U. 8. Nat. Mus., VI, pp. 152-154. INDEX. [The black-faced numbers indicate specific headings. ] A. : Page ANG TO ink Che Cee See ees 15 Abundances]. eee 190 Agknowiedements = —— === Se tf Adams, C. C_-- 2, 7, 187, 189, 190, 191, 193 Affinities (see also Description of CxGheiorm() y= =—— a 2, Goa, O40 LOO PCR pis pA ete ee SE Se ee See 188, 193 Analeplates sa 25 Seo 22s bo Ee 8,9 ANCUSULPOS ERIS = 52 —= = = 9, 25,735, 36, 40) 43, 120, 125, 127, 187, 146, 157, 158, 189 AT KIiM SOM). At Ses = US See eS 112 PMOTMATCHUS§ = Bese ts SS 9, 120 Arata =o f= Ss 90, 188, 147, 148, 156 PAGINA a re Se a I oe 158, 164 B. Baile yas Vee ee 99, 100, 160, 162 SOG ats ee ee eee es il Baird) Ss. heand. Girard == sss 8, 35, 58, 66, 70, 145, 147, 176 Bakers he Casta eS ee 45 Baronis-mulleri=!= 2222S oe 124 Bipliocniap hiv sa -—= = Se 195 Biscutata ____ 138, 139, 144, 147, 148, 155 POGOe ee ee Kaloo al Gs Boulenger, G. A__ 5, 100, 128, 161, 162, 165 Brachystoma— 2 — = Os SOs) 950904596 Branson sh) ph <= ee. ae 75, 100, 141 PS Talvane Wit dam ee a Se 61, 99 Breeding season. = 8 oS 13 Brood. (See Productiveness. ) IBTOUS ie ae eae ee ee ee 71 Brown, A. Hos= 2-22-25 1, 5, 8, 35, 39, 54, 58, 61, 86, 88, 92, 144, 149, 156, 186, 190 Brunnea= 222 SS ee aoe 138, 145 BUMpusi ees ce se 2 ee 8 lta rile 2a ee 4,5, 9, 11, 145 1, 20; 27; 28, 31, 35, 37, 38,40; 43, 44, 85, 86, 87, 129, 185, 186, 189, 191, 194 (Ch Center of origin. (See Origins.) Cephalicn plates — === Bes So 8 Characters used in systematic C7 se 4, 7,15, 40, 41 Clap ll toyota ee 9, 120 Chin tS biel ds eee ee ee 9, 15, 38 Chrysocephalla == ae 164, 165 (Oi ire tied 6 ew | Oe OS es ee ee 89 Page. Clements, F. E., and Pound, R. (See Pound and Clements. ) Colton se ee ee ee 13 C@olevely cS ase bs 2 Se See 98 Collarisr soe Dae eee eee 158, 159, 163, 184 Color (see also Description of EACH SLOT) ieeeene eee ee ee eolO net Coluber = 2— 4 ae ee 8, 148, 158, 176 GConcinnus|s==--— hee ae ee 5, 6, 34, 36, 38, 41, 42, 44, 171, 178, 186, 191 Convergence. (See Parallelism.) Cooper sin Guana eee ee 173 COopeninws = ee be ee eee ee 147 Cope Dea 1, 4, 5, 6, 35, 45, 49, 54, 56, 61, 63, 66, 80, 85, 86, 87, 88, 94, 98, 121, 124, 130, 131, 138, 144, 145, 162, 164165, G6) 17s 2, 7G; 185.8190 Couchia 2 ae ee 1338, 145 Coues, E_____ 46, 47, 72, 75, 78, 85, 138, 142 Coues, E., and Yarrow, H. C_ 37, 74, 85, 146 Cox4h 2s ee eee ae 181 CyClides 32s = ee ees | See 158, 164 Cyrtopsis=.2— 2. =. See eee 132, 158 D. Darwin Cs Ss). see 187 Definite variation (see also Ortho- SENESIS)| = oe Le ee 5, 23, 193 Development, parallel. (See Paral- lelism. ) Diagrams. (See Statistical meth- ods. ) Distribution (see also Description of each form) — 12, 42, 187, 189, 191, 192 Ditmas seh == 90, 99, 108, 112, 179 Divisions. (See Primary divisions. ) Dorsalis-=—— === = 158, 159, 166, 168, 172 Dorsal scale rows (see also Descrip- ion Olea ch eLOLi) =e a 4, 9, 15, 16, 36, 37, 40, 41 Dorsal stripe (see also Description OLmeay Che LOL) mes ee ial Se Tyra TN eee es ee, ee 2021 Loe Silty -S2) Vols s04,, 9%. 119s 220 295 1525, Tht, 158, 1885" 191, 193, 194, 295 BE. ileoans aaa =e See AS19512. 17, 24, 31, 34, 36, 41, 42, 43, 58, 61, 62, 66, 67, 68, 74, 86, 120, 130, 135, 137, 188, 147, 148, 149, 150, 152, 155, 156, 157, 158, 195 199 200 INDEX. Page. L. Page. Mlecansie ero po aaa aa ee 41, 2, 66, 67, 68, 120, 130, 157, 158,195 | Labials (see also Description of HUiCouiee lee ee Se 13 each form)/=2=—-=--—= O11 5: 22) 4074s DOES SO 17, | Lateral spots. (See Spots.) 18, 33, 34, 38, 41, 43, 62, 67, 86, 132, 158, Lateral stripe (see also Description 165, 166, 168, 171, 172, 173, 186, 187, 189 Butaenia (see also Synonymy of each TOR) a a eee Se 8 Leber bly asic £00 aaa a eee eg 14 Bye, relative position of_______-_ 9, 25, 39 ie Haineyie-— 2 > 98 Kitzinger, Tie Jos >= = 8 HlawlabrisS ==. eee 44, 49, 56, 57 Food (see also Description of each FOR) 2 Se ee ee 12 BeLnms -— === = Ses Se ee 3, 6, 39, 40, 41, 43, 188, 189, 190, 191, 192, 194 Wowler:} Els We 2- ee Se ee 93, 95 Brontgll 3 8 eos ee ee ee 8, 39 G. @armen Wel ae Sa ae 100, 168, 177 Garments See ee ee al tay, (oh 53 Geneials. (See Chin shields.) Geographic variation. (See Variation.) Girard, C., and Baird, S. F. (See Baird and Girard.) ; Godmamniya== 2 es a ase 164, 165 Gramineae 222 = — __ eae eee 176, 184, 185 Grecarious, tendency —--=—--— === — 13 Grinnell, J. and Grinnell, H. W__ 133, 168 Groups of garter snakes (see also Radix, Sauritus, Elegans, and Sir- CAlisiyeroups) io — 39, 40, 41, 42, 187 Gunther, -A. Cosi Ge==—— = 100, 125, 165 els Habitat. (See Habits.) Habits (see also Description of each OTN) ee ee 7, 12, 190, 192 Ea Lowe liege) 222 > ieee ee eS 173, 176 Hammondi____ 8, 17, 34, 36, 40, 42, 43, 66, 67, 120, 124, 133, 145, 146, 147, 157, 158 ay Oe a ee 13, 14, 35, 80, 88 Haydenis 2.8321 ee 70, 84, 85 teil prin,, Avt- = + 46 enshaiwi === 2s ee eee 138, 146 Eijbernationy.~ 22 0= 2s: =e 13,179 nN ORs Sa a ee eee 60, 61 Hyde, Ja---<. ==. 191 I: Infermalisicoss=-- 5 = S= 138, 147, 148, 156 Infralabials. (See Labials.) Iusigniasvam, 2=—— === —=— 44, 45, 54, 56, 57 imtérnasals)--=- =o. -—- = s-e 8 Ksolation.2——— = = 6, 193 Ve Jordan, Di Si) eee 192 Ke Keunicott, B22 44, 53, 84, 133, 145 Key -to: Garter snakes=— 43 Of each) form) 22a 4, 11, 35, 40, 41, 42 Leptocephaluse= = 5, 138, 147, 148 Lineolata. == Fen ast ee See 138, 145 Gaborone, (endl yori ee ata Palys, orea hess — = a ee 8 M. Macrostemma ___ 44, 45, 54, 56, 57, 121, 125 Maps = se ee 8 Marcianus __-_ 11, 34, 35, 37, 38, 40, 42, 43, 44, 58, 70, 71, 78, 86, 96, 97, 106, 147 Matertalexamined 222-2 ee 7 Meéelit Sh Fh es 2oe 2 eee 8, 158 Megalops___ 17, 18, 25, 26, 29, 34, 36, 37, 38, 40, 43, 44, 58, 68, 69, 70, 71, 86, 87, 96, O97, 106; 10%, 1325 189, 195: Miclanisme’ === sa =2 SS = 34, 38, 57, 145 Melanogaster___ 9, 11, 12, 34, 38, 41, 44, 45, 1205 12 123) 124, SO Si. loners Melanotaenis. 2s 2-2 eee 70, 80 Miéldola, Re=2- 23222 eS ee _ 2 Mentaliplate. “= ==> Ss. = eee 9 Merriams (C) (hark) ee 138 Mesomeladena==— 32s ee 124 Mesomelanus==—=2 5 - = e . 124 Methods. (See Taxonomy.) IMOTSE Mi 22 = ee ee et 88, 92, 185 Mialtimacuilatts| 2 = = 120, 123 Mutations;-=2 ==... = a eee 194 N. Nasal. 24 2 2 a ee eee 8 Natricidse: =. 2222 2. '= 32-22. 8 Nev tring 682 2 ee 8 Nigrolateris == 2.22. ee ees 58, 64 INostrilc: 4.28. oo a eee 8 oO. Obscura. 22 =s322-— == 87, 88, 92, 176, 185 Occipitals-——==— =. = Ses 2 ee 8 Ocellata 22s 22s = eee 158, 159, 163, 171 Oculars. (See Preocular and Post- ocular.) Olympia = 2 ee 147, 156 Ordinatis;— ==-——- = = = 176, 182, 184, 185 Ordinoitdes 42. = eee 5, 6, 8) 95 17, 22, 24, 26, 27, 31, 34, 36, 38, 41, 42, 44, 90, 120, 144, 147, 157, 158, 176, 189, 194 Origins. =—_= 2 ee aya las ye akeiry Ornate sos oe ee 158, 159 Orthogenesis (see also Definite varia- tion) 2222S) = Se 194 Orimant A w= ee 192 P. Palpechinolds == - =2~—- == 21 Palliduls<. 2222.4 =e 176, 184, 185 Palmén 2220) 2 eee 187 Parallelism s3s22.———-— == 5, 158, 188, 189 INDEX. 201 Page. Page. Parietalis_____ AGMA SO oy OS, 41, 425 | Sines == 2-2 — 5; Gyala elas 44, 66, 67, 71, 74, 85, 147, 158, 164, 166, 15 1G is, hoy 2D, 27, So, ah 36, 38, 41, 178, 174, 175, 176; 178, 181, 184, 185, 186 PRG Olle ne oe ke. 14 iBeariitayimmo nde sess = 2 8 Peters, WwW i ee ee ee eee 124 JERSE oy oo ee eee ee 33,480, 41, 44, 120; 127, 1380, 157, 158 Rickeningiese= — = 422i ese Delis noel Pihwitioniars= saa ee 138, 146 Postoculars (see also Description of Gacr atone. 2 Le ae 9 Found, R., and Clements, F. E._____ 79 FEB OCU SUIS See ye an ee i PeRennontailgies = ee ts Pe Se 8 Preoculars (see also Description of CHCHELOTMN) (esse 8, 32, 40, 41, 42 Primary divisions of garter snakes_ 40, 41 Productiveness (see also Description Oamea Cor foci) == == = = t= 14, 191, 194 EONS ee =e es 18, DOW OL oe, OOF Ol; oS, 40) 435 86, 8, SS tl ee LS Oe 221 GO5 189 Vulchrilatus 158, 162, 166 LSE [bc ee 4, 6, 10, SA ig NO LO, le 265 2 28, ol, 32, 34, 35, 37, 38, 40, 43, 44, 61, 69, 70, 86, 87, 93, 94, 96, 106, 186, 191, 194, 195 Radix sroupss—=———— 40, 42, 44, 58, 68, 69, 96, 97, 106,-107, 120, 1380, 132, 157, 195 Range. (See Distribution.) VCduTGks an Gee ae = sees a tee a 90 Ofna See aes oe ee Se eee eS 124 VaR) Bas | Soe Se re ee eee 8 Relationships (see also Affinities) —_ 4 TOYS EEE) | 2x Aas he a eee ae ee 8 UCI SES ON ali pe een ee eee TES ey 147, 158 RuLopunctatum— === sess UPTO. pleats P53 Rachivens A Ge s=== 28, 46, 59, 63, 88, 89, 92, 96, 104, 141, 160, 161, 162, 168, 190 PEUUIT LOM S fee ee ee 98, 104 S. Nackenisee Ss 2 ae 4 Sot. 40.43: 987.09 7 LOGO 7, dale ial A AS ALO SAUTMS 2 ele ee Shel, 18 22. 23, 32, 34, 35, 36, 31, 68, 40, 43, 86:88: 90, 98, 99, 106, 107, 108, 112, 119, 121 Sauritus group -_-- 10, 32, 40, 42, 98, 110, 111, 119, 180, 1382, 156, 157, 165, 189, 195 SGI Rn 3 eS ee eee UO ir a8} Bk 414s ON ee tod. a2) 157. The) 189 PeSeUED [OS ee eS 8 Scale rows. (See Dorsal scale rows, and ventral and subcaudal plates. ) SCR One ee et eer Ee 127 Stole es Od [ee SS ee 14 Scutellation (see also Dorsal scale rows, ventral and _ subcaudal (LOUCN)) =e ee Ge 8, 16, 188, 191, 193, 194 Semitasciata..= 2 = see 171, 176, 184 Serpentes, systematic work on_____ 1 Sexual variation “-2=s-———2-—2= 15, 21.23 O 42, 44, 66, 74, 86, 87, 88, 90, 92, 96, 125. 158, 168, 171, 172, 173, 176, 186, 190, 191 Sirtalisteroupe—————— = === Ss - 41, 42, 130; 1382; 157, 158, 186, 189, 191, 195 Size (see also Dwarfing and Descrip- ELON) Of Miho nM) pee ee 21 Smith's Wiles eer 2 eee Sa 8 SO Spots (see also Description of each LOrM) a eee LOM oo tO! AN, 42 Deatisticalys methods] =e aaa = 8 Sleeres “pps + = eee 2 Se or se eee 192 Stejnevenw lh 5-2 e ee TaD SS SCOR CW pen ss ee ee 8, 89, 93, 192 Stripes (see also Dorsal stripe and lateral Stripes oe = eee a) Ses 7,15 SYDOCEMNUIS S= == == eee ae ee 9 Subcaudals (see also Description of Ca chestorm)) 2 22 eee ee ye kaye) Sumichrast. hs. {tS eee ee 128 Sumichrasti_11, 34, 41, 44, 158, 164, 186, 189 Supralabials. (See Labials.) Surface WH. Avan SEE th eee 112 Synonymy (see also Description of each form) == sa ae eee ee 39 Systematic work. (See Taxonomy.) Abe Tail length (see also Description of eachistorm)=——=—— 7, 15, 32, 37, 40, 41, 42 Waxonomy,=2= == a ‘* ; , “! « ia Pi oy +e ~ 4 Sen — a ’ = /¢ ty tl! = 2a? at ee Jr ee eS es a ie way va hie, he ie (hr Bits 7 iu Dal iy oy, a ane | } | | at ’ ‘ i & 4 ie . Ly ae ry eae ih : ‘ if ay uf bi men A ae ; SMITHSONIAN LE A i Variations and genetic relationships of